GKSS-2008-12 [9,8 MB] (PDF)

ISSN

034

4-9

629

GKSS 2008/12

Author:

G. A. Pinheiro

Local Reinforcement of Magnesium Componentsby Friction Processing: Determination of BondingMechanisms and Assessment of Joint Properties(Vom Promotionausschuss der Technischen Universität Hamburg-Harburgals Dissertation angenommene Arbeit)

Local Reinforcement of Magnesium Componentsby Friction Processing: Determination of BondingMechanisms and Assessment of Joint Properties(Vom Promotionausschuss der Technischen Universität Hamburg-Harburg

als Dissertation angenommene Arbeit)

GKSS-Forschungszentrum Geesthacht GmbH • Geesthacht • 2008

Author:

G. A. Pinheiro(Institute of Materials Research)

GKSS 2008/12

Die Berichte der GKSS werden kostenlos abgegeben.The delivery of the GKSS reports is free of charge.

Anforderungen/Requests:

GKSS-Forschungszentrum Geesthacht GmbHBibliothek/LibraryPostfach 11 6021494 GeesthachtGermany

Fax.: +49 4152 87-1717

Als Manuskript vervielfältigt.Für diesen Bericht behalten wir uns alle Rechte vor.

ISSN 0344-9629

GKSS-Forschungszentrum Geesthacht GmbH · Telefon (04152) 87-0Max-Planck-Straße 1 · 21502 Geesthacht / Postfach 11 60 · 21494 Geesthacht

GKSS 2008/12

Local Reinforcement of Magnesium Components by Friction Processing: Determinationof Bonding Mechanisms and Assessment of Joint Properties

(Vom Promotionausschuss der Technischen Universität Hamburg-Harburg als Dissertation angenommene Arbeit)

Gustavo Alves Pinheiro

189 pages with 103 figures and 32 tables

Abstract

The development of new creep-resistant and cost effective die casting magnesium alloys, such as AE,MRI, MEZ, ACM, AXJ, AJ, WE, have emerged as an alternative, to fulfil the modern demands in struc-turally relevant applications, such as engine blocks, gears and converter boxes. However, in most cases,magnesium components are screwed with aluminium and steel bolts, which lead the screwed joints tolose the preload force, due to relaxation. This barrier thereby limits the broad use of magnesium withinthis segment and should somehow find an adequate solution to help overcome this limitation. Further-more, together with alloy development and the addition of reinforcement (MMCs), local materialengineering processes have been conceived and are considered a method to improve the properties andtherefore expand the number of potential applications for magnesium alloys. In this context, FrictionWelding (FW) and particularly Friction Hydro Pillar Processing (FHPP), which can be described as a drilland fill process, appear to be an alternative to make the use of magnesium more widespread. For thisreason, FHPP is intended to be used, to locally reinforce the mechanically fastened magnesium compo-nents. With this approach, regions submitted to the stresses imposed by tightening forces can becompensated by the use of a material with superior properties. It is not required to fabricate the wholestructure from an expensive material, thus saving costs and thereby satisfying the economic pressuresof an increasingly competitive global market.

In the present work, a preliminary experimental matrix was defined and used to determine the optimalwelding conditions for each specific material combination selected. Further, elaborate experimental tech-niques are used to describe the process parameters-microstructure-properties relationships and theconsequent mechanisms leading to bonding in FHPP welds in similar and dissimilar configurations. Thewelds were performed using a hydraulic powered friction welding machine, originally designed and builtas a portable stud welding unit, delivering up to 40 kN welding force and 8000 rpm. All welds were mon-itored, analysed and evaluated using a purpose-built data recording system. AZ91, AE42 and MRI230Dmagnesium grades were used in the experimental programme. The results obtained in the course of thisstudy have shown the feasibility of FHPP to produce high strength welds with mechanical propertiescomparable to those of the base material. Defects, such as porosity or lack of bonding, were notobserved. Furthermore, the welding pressure rather than the upsetting was found to have a major influ-ence in the final weldment. The influence of process parameters on heat generation and bonding qualityis similar to that known from rotational friction welding. It could be demonstrated that for dissimilarMRI230D to AZ91D and for similar AZ91D to AZ91D welds, the consumable member is fully plasticisedacross the bore of the hole and throughout the thickness of the workpiece. However, for AE42 to AZ91Dwelds, the stud was not completely plasticised across the bore of the hole and significant microstructuralchanges were restricted to a narrow area around the bonding line. Hardness profiles indicate a substan-tial reduction in scattering, as soon as the stud material is reached. Hardening or softening phenomenawere not observed. Transverse tensile and pull-out testing confirmed the feasibility of the process to pro-duce high strength welds, with failures taking place outside the welded area in most of the cases. Jointperformance, in terms of creep and bolt load retention, was also tested and showed promising results.Although creep properties were demonstrated to be inferior within the extruded material, in comparisonwith the base materials (BM), the creep resistances of reinforced samples were always superior to thoseof purely unreinforced AZ91D-T6. The strength of the reinforced joint in both tensile and compressiveBLR tests was elevated relative to the BM, achieving in some cases 93 % of BM values.

Lokale Verstärkung von Magnesiumkomponenten mittels „Friction Hydro Pillar Processing”:Bestimmung der Bindemechanismen und Bewertung der Verbindungseigenschaften

Zusammenfassung

Die Entwicklung neuer kriechbeständiger und kostengünstiger Magnesiumdruckgusslegierungen wie AE,MRI, MEZ, ACM, AXJ, AJ und WE hat sich als Alternative zur Erfüllung der tatsächlichen Forderungender strukturell relevanten Anwendungen, wie bei Motorblöcken und Getrieben, erwiesen. Allerdings sindMagnesiumbauteile in den meisten Fällen mit Aluminium- und Stahlschrauben zu verschrauben. Dieunterschiedlichen thermischen Eigenschaften führen bei den betriebsbedingten Lastwechseln dazu,dass die verschraubte Verbindung aufgrund der Relaxation die Vorspannkraft verliert. Diese Problematikbegrenzt damit die Verbreitung von Magnesiumlegierungen in diesem Segment. Legierungsentwicklungund Partikelverstärkung (MMCs) sowie lokale Werkstoffprozesse wurden in letzter Zeit als Methoden zurVerbesserung der Eigenschaften entwickelt. In diesem Zusammenhang stellt sich das Reibschweißenbesonders in Form des Friction Hydro Pillar Processing (FHPP) als eine der modernsten und geeignetestenOption vor. In diesem Schweißverfahren wird ein Bolzen unter Rotation in eine Bohrung eingeführt. Ander Kontaktstelle wird aufgrund des aufgebrachten hohen axialen Drucks Reibungshitze generiert. DasBolzenmaterial wird dadurch plastifiziert und durch die örtlichen hohen Temperaturen eine metallischeVerbindung zum Grundmaterial hergestellt. Das Ziel dieser hier dargestellten Arbeit ist die lokale Ver-stärkung von Magnesiumbauteilen mittels Einbringen einer festeren Magnesiumlegierung (GKSS-Patent n°. 816 p Gk 3.03). Durch diesen Ansatz können die einwirkenden Kräfte im beanspruchten Bereichdurch die Einbringung eines Werkstoffs mit festeren mechanischen Eigenschaften besser aufgenommenwerden. Einerseits muss die ganze Struktur nicht aus einem teuren Material hergestellt werden, undandererseits können Kosten eingespart werden, um sich dem wirtschaftlichen Druck eines zunehmendwettbewerbsorientierten globalen Marktes anzupassen.

In der vorliegenden Arbeit wurde in erster Linie ein Schweißparameterbereich definiert und untersucht,um die optimalen Schweißbedingungen für jede spezifische Materialkombination festzustellen. Außerdemwurden experimentelle Techniken zur Beschreibung der Korrelationen zwischen den Prozessparametern,Mikrostruktur sowie der daraus resultierenden mechanischen Eigenschaften und den zugrunde liegendenBindemechanismen in FHPP-Schweißnähten in artgleichen und -ungleichen Konfigurationen. DieSchweißungen wurden mit einer hydraulischen Reibschweißmaschine, welche ursprünglich als portableBolzenschweißeinheit ausgelegt wurde, durchgeführt. Sie generiert eine axiale Schweißkraft bis zu 40 kNbei einer Drehzahl bis zu 8000 U/min. Alle Schweißungen wurden mit einem speziell dafür entwickeltenDatenaufzeichnungsgerät überwacht, analysiert und bewertet. Die Magnesiumlegierungen AZ91, AE42und MRI230D wurden in dem experimentellen Programm untersucht. Die Ergebnisse, die im Verlaufdieser Studie gewonnen wurden, haben gezeigt, dass mittels FHPP hochfeste Schweißungen mit mecha-nischen Eigenschaften, vergleichbar denen des höherfesten Grundwerkstoffs produziert werden können.Schweißfehler, wie Porositäten oder Bindedefekte, konnten nicht beobachtet werden. Die axialeSchweißkraft weist dabei einen wesentlich größeren Einfluss als die Gesamtverkürzung auf die Qualitätder Schweißung auf. Der Einfluss von Prozessparametern auf die Wärmeerzeugung und Bindungs-eigenschaften erwies sich als vergleichbar zu den beim Rotations-Reibschweißen beobachteten Effekten.Für artungleiche Verbindungen MRI230D auf AZ91D und für artgleiche AZ91D-Verbindungen konnte dievollständige Plastifizierung des Verstärkungsmaterials über den gesamten Verbindungsquerschnittnachgewiesen werden. Im Gegensatz dazu wurde für die artungleiche Verbindung AE42 auf AZ91Dkeine vollständige Plastifizierung erreicht. In diesem Fall konnten lediglich in einem engen Bereich um dieBindelinie deutliche mikrostrukturelle Veränderungen beobachtet werden. Härteprofile zeigen eine deutlicheReduzierung der Streuung, sobald das extrudierte Material erreicht wird. Härt- oder Erweichungs-phänomene wurden nicht festgestellt. Querzug- und Pull-out-Tests bestätigten die Tauglichkeit desVerfahrens zur Herstellung von Schweißverbindungen hoher Festigkeit. Das Versagen erfolgte in denmeisten Fällen außerhalb des geschweißten Bereiches. Die Belastbarkeit der Verbindungen in Bezug aufKriech- und Spannungsrelaxationsbeständigkeit wurde getestet und zeigte viel versprechende Ergebnisse.Obwohl die Kriecheigenschaften der extrudierten Materialien nicht die der Ausgangswerkstoffe erreichten,konnte ein deutlich erhöhter Kriechwiderstand der verstärkten Proben gegenüber dem reinen unver-stärkten AZ91D-T6-Grundmaterial nachgewiesen werden. Die Festigkeit der verstärkten Verbindungen,relativ zu den Grundwerkstoffen während des Spannungsrelaxationstests sowohl unter Zug- als auch Druck-belastung, erreichte in einigen Fällen 93 % der Werte der Grundwerkstoffe der Verstärkungsmaterialien.

Manuscript received / Manuskripteingang in TKP: 12. Dezember 2008

1 Introduction...................................................................................................... 1-1

2 State of the Art................................................................................................. 2-5

2.1 General Metal Characteristics .................................................................... 2-5

2.2 Nomenclature and Alloying Elements ........................................................ 2-7

2.3 Generalities, Phases and Structures of Mg and its Alloys.......................... 2-9

2.3.1 Magnesium-Aluminium-Zinc (AZ) - Casting Alloys................................ 2-9

2.3.2 Magnesium-Aluminium-Rare Earths (AE) - Casting Alloys ................. 2-11

2.3.3 Magnesium-Aluminium-Calcium-Strontium (AXJ) - Casting Alloys ..... 2-12

2.3.4 Combined Properties .......................................................................... 2-12

2.4 Bolted Magnesium Joints ......................................................................... 2-14

2.5 Friction Welding – a Literature Review..................................................... 2-18

2.5.1 Relevant Process Parameters ............................................................ 2-20

2.5.1.1 Relative Speed of the Faying Surfaces ......................................... 2-21

2.5.1.2 Normal Force (Axial Pressure) ...................................................... 2-22

2.5.1.3 Heating Time (HT) ......................................................................... 2-24

2.5.1.4 Upsetting (Axial shortening)/Upsetting Rate.................................. 2-25

2.5.2 Process Characterisation – Phases of the Process ............................ 2-26

2.5.2.1 Phase 1 – Rubbing Phase............................................................. 2-27

2.5.2.2 Phase 2 – Heating Phase.............................................................. 2-29

2.5.2.3 Phase 3 – Braking Phase .............................................................. 2-30

2.5.2.4 Phase 4 – Bonding Phase ............................................................. 2-31

2.6 Friction Hydro Pillar Processing ............................................................... 2-32

2.7 Friction Welding in Magnesium and its Alloys .......................................... 2-38

3 Experimental.................................................................................................. 3-42

3.1 Base Materials ......................................................................................... 3-44

3.2 Studs and Holes Configuration................................................................. 3-44

3.3 Welding System ....................................................................................... 3-46

3.4 Data Acquisition System (DAS)................................................................ 3-48

3.4.1 Data Recording System – DRS/Process Stability ............................... 3-48

3.4.2 Temperature Measurements............................................................... 3-48

3.5 Testing Equipment and Procedures ......................................................... 3-49

3.5.1 Welding Procedure ............................................................................. 3-49

3.5.2 Metallurgical Characterisation............................................................. 3-49

3.5.2.1 Optical Microscopy ........................................................................ 3-49

3.5.2.2 Grain Size...................................................................................... 3-50

3.5.2.3 Scanning Electron Microscopy/EDS.............................................. 3-51

3.5.3 Hardness Testing................................................................................ 3-51

3.5.4 Pull-out Testing................................................................................... 3-52

3.5.5 Transversal Tensile Testing................................................................ 3-54

3.5.6 Bolt-Load Retention Testing (BLR) ..................................................... 3-55

3.5.7 Creep Testing ..................................................................................... 3-56

4 Results .......................................................................................................... 4-58

4.1 Metallurgical and Mechanical CharacteriSation of Base Materials.......... 4-58

4.1.1 Base Material AZ91D-T6 .................................................................... 4-58

4.1.2 Base Material AE42 ............................................................................ 4-61

4.1.3 Base Material MRI230D...................................................................... 4-63

4.2 Similar AZ91D-T6 joints – 6 mm diameter/12 mm upsetting .................... 4-67

4.2.1 Process Monitoring ............................................................................. 4-69

4.2.2 Thermal Cycle..................................................................................... 4-71


4.2.4 Mechanical Characterisation............................................................... 4-80

4.2.4.1 Hardness Tests ............................................................................. 4-80

4.2.4.2 Pull-out Tests ................................................................................ 4-81

4.2.4.3 Transversal Tensile Tests ............................................................. 4-82

4.3 Dissimilar Joints – 8 mm Diameter/10 mm and 20 mm Upsetting ............ 4-83

4.3.1 Process / Temperature Monitoring...................................................... 4-84


4.3.2.1 AEAZ Joints................................................................................... 4-87

4.3.2.2 MRIAZ Joints................................................................................. 4-90

4.3.2.3 AZAZ Joints................................................................................... 4-93

4.3.3 Mechanical Characterisation............................................................... 4-96

4.3.3.1 Hardness Tests ............................................................................. 4-96

4.3.3.2 Pull-out Tests ................................................................................ 4-99

4.3.3.3 Transversal Tensile Tests ........................................................... 4-100

4.4 Joint Performance .................................................................................. 4-102

4.4.1 Creep Tests ...................................................................................... 4-102

4.4.1.1 AE42 Studs (Reinforcement)....................................................... 4-102

4.4.1.2 MRI230D Studs (Reinforcement) ................................................ 4-103

4.4.1.3 AZ91D-T6 Studs (Reinforcement) ............................................... 4-103

4.4.2 Bolt Load Retention Tests................................................................. 4-104

4.4.2.1 BLR Tests – In Tension ............................................................... 4-105

4.4.2.2 BLR Tests – In Compression....................................................... 4-107

5 Discussion of Results .................................................................................. 5-110

5.1 Base Material CharacteriSation.............................................................. 5-110

5.1.1 Mg-Al-Zn [AZ] – Ingot (heat treated to T6 condition)......................... 5-110

5.1.2 Mg-Al-Rare Earths [AE] – Ingot (cast condition) ............................... 5-110

5.1.3 Mg-Al-Ca-Sr [AXJ (MRI)] – Ingot (cast condition) ............................. 5-111

5.2 Similar AZ91D-T6 Joints – 6 mm Diameter/12 mm Upsetting ................ 5-113

5.2.1 Process and Thermal History............................................................ 5-113

5.2.2 Metallurgical Characterisation / Joint Formation Mechanisms.......... 5-121

5.3 Dissimilar Joints – 8 mm Diameter/10 mm and 20 mm Upsetting .......... 5-127

5.3.1 Process and Thermal History............................................................ 5-127

5.3.2 Metallurgical Characterisation / Joint Formation Mechanisms.......... 5-132

5.3.2.1 AEAZ Joints................................................................................. 5-132

5.3.2.2 MRIAZ Joints............................................................................... 5-137

5.3.2.3 AZAZ Joints................................................................................. 5-141

5.4 Mechanical CharacteriSation ................................................................. 5-143

5.4.1 Hardness Tests................................................................................. 5-143

5.4.2 Pull-out Tests.................................................................................... 5-145

5.4.3 Transversal Tensile Tests................................................................. 5-146

5.5 Joint Performance .................................................................................. 5-150

5.5.1 Creep Tests ...................................................................................... 5-150

5.5.2 BLR tests .......................................................................................... 5-160

6 Summary and Conclusions.......................................................................... 6-163

7 Suggestions for Further Work...................................................................... 7-169

8 References .................................................................................................. 8-170

List of Symbols and Abbreviations

AZAZ-1C Weld designated to metallurgical characterisation according to condition

“1” described in Table 4.10





AZAZ-06/12 Series of joints welded using 06 mm studs/holes (tip and bottom hole

diameter respectively) and 12 mm upsetting





AEAZ Dissimilar joints using AE42 studs welded with AZ91D-T6 base plates

(8 mm stud tip diameter / 10 mm and 20 mm upsetting)

AZAZ Similar joints using AZ91D-T6 studs welded with AZ91D-T6 base plates

(8 mm stud tip diameter / 10 mm and 20 mm upsetting)

MRIAZ Dissimilar joints using MRI230D studs welded with AZ91D-T6 base

plates (8 mm stud tip diameter / 10 mm and 20 mm upsetting)

AWS American Welding Society

BL Bonding Line

BP Base Plate - The lower stationary AZ91D-T6 workpiece

BLR Bolt Load Retention (Test)

BM Base Material

DAS Data Acquisition System

DRX Dynamic Recrystallisation

EDS Energy dispersive spectroscopy

FP Forging Pressure

HAZ Heat Affected Zone

R Radius of the workpiece/stud

RS Rotational Speed

RZ Recrystallised Zone

TMAZ Thermo-mechanical Affected Zone

UPS Upsetting

WP Welding Pressure

1-1

1 INTRODUCTION

Since after the Second World War, when its consumption slumped from 228,000 to

10,000 tons per annum, magnesium again appears strongly placed among the new

innovative materials [1]. In recent years, magnesium alloys have gained increasing

interest in different branches of industry. This interest can be attributed to the fact

that the material has overcome some of its old barriers, as the fairly high base

material prices somehow limited to a few technical alloys and associated with some

costly recycling possibilities [2,3]. Over recent years the industrial output of

magnesium alloys has been rising almost 20 % per year [4]. This increasing

application of lightweight alloys particularly matches the interests of the automotive

industry, with its goal of reducing the weight of vehicles to make them more fuel

efficient [5]. In this context, several new Mg alloys for different applications have been

developed recently, with the intention of obtaining an adequate combination of

various properties. Evidently, a perfect combination of creep resistance, castability,

mechanical properties, corrosion resistance and affordable cost is rarely achieved

and most of the alloys can only address some of the required properties and

performance.

The shift from air-cooled to water-cooled engines in the 1970s dramatically reduced

the use of magnesium alloys. The existing AZ and AM series are not suitable for

manufacturing drive train parts, which operate at temperatures above 130 °C.

Aluminium-silicon and aluminium-rare-earths based creep resistant alloys developed

in the 1970s and 80s also have not found their place in powertrain applications, for

various reasons such as poor castability and corrosion resistance, increased costs

and low strength. However, the development of new creep resistant and cost

effective die cast magnesium alloys, primarily based on the use of Al, RE, Mn, Ca,

Sr, Zr as alloying elements, has emerged as an alternative to fulfil modern demands

in other structurally relevant applications, such as engine blocks, gears and converter

boxes [6]. Consequently, more and more components have been fabricated using

magnesium and its alloys. Table 1.1 and Table 1.2 list the present situation

concerning the application of magnesium within the automotive industry, whereas

Table 1.3 estimates the foreseen consumption of magnesium in 2015.

1-2

Table 1.1: Current applications in the automotive industry.

INTERIOR EXTERIOR Seat and components Engine hood

Instrument panel and beams Roof panels

Knee bolsters Rear deck lid

Pedal brackets Radiator support (1)

Air bag retainers Grill opening support

Sunroof components ENGINE DRIVE TRAIN Mirror brackets Cylinder head covers Manual transmission housing

Headlight retainers Intake manifold 4WD transfer case

Inner door frames Engine block Automatic transmission

Support pillars Oil pan

Steering column components Starter-Alternator housing

(1) Highlighted columns are under development

Table 1.2: Vehicle platforms with high magnesium content [4,7].

Vehicle Platform Content per vehicle Volvo LCP 2000 49kg

Ford PNGV P2000 38kg

GM full size vans (Savana/Express) Up to 26.3kg

Daimler Benz SL 17 to 23kg

GM minivans (Safari e Astro) Up to 16.7kg

Ford F-150 truck 14.9kg

VW Passat, Audi A6/A4 13.6kg to 14.5kg

Porsche Boxter Roadster 9.9kg

Buick Park Avenue 9.5kg

Alfa Romeo 156 9.3kg

Daimler Benz SLK Roadster 7.7kg

Chrysler minivans 5.8kg

Table 1.3: Magnesium use forecast in the automotive industry [8].

2000 2005 2015 140,000 tons 240,000 tons 600,000 tons

In most of the cases, magnesium alloys need to be mechanically fastened, not only

in a similar, but sometimes in a dissimilar configuration, by means of screws of

different materials. In consideration of the elevated operating temperatures

experienced by engine blocks and powertrain components, the creep resistance of

magnesium alloys remains a major issue for such applications. In the literature, it has

been reported that the use of aluminium and steel screws in magnesium components

lead the screwed joint to lose the preload force, due to relaxation [9,10]. This barrier

thus limits the broad use of magnesium within this segment and needs somehow to

find an adequate solution to help overcome this limitation.

1-3

Furthermore, together with alloy development and the addition of reinforcement

(MMCs), local material engineering processes have been conceived and are

considered a good method to improve the properties and therefore expand the

number of potential applications for magnesium alloys. In this context, Friction

Welding (FW) and particularly Friction Hydro Pillar Processing (FHPP), which can be

described as a drill and fill process, appear to be an alternative to make the use of

magnesium more widespread. Although specialised literature involving FW on

magnesium alloys has not expressly contributed to the development of this work and

no literature on FHPP of lightweight alloys has been found, FHPP is intended to be

used to locally reinforce mechanical fastened magnesium components. With this

approach, the regions submitted to the stresses imposed by the tightening forces can

be compensated by the use of a material with superior properties. It is therefore not

necessary to fabricate the whole structure from an expensive material, thereby

saving costs and addressing the economic pressures of an increasingly competitive

global market.

Due to the current problems involving Mg bolted joints, there is an evident need for

research into alternative solutions. Hence the overall objective is to locally reinforce

Mg components through the welding of high creep resistance inserts (studs) to be

used in automotive applications. Within this overall objective, some specific aims can

be defined as follows:

To describe the bonding formation mechanisms in FHPP of Mg alloys, based

on the behaviour of different material combinations and welding parameters,

leading to an understanding of the microstructural changes during welding;

To characterise the base materials (BMs) in terms of structures and

properties, in order to compare and evaluate the local reinforcement

properties based on the base material’s mechanical behaviour. Further, a

correlation of BM microstructure/process/final microstructure is also intended

to be done;

1-4

To determine the local mechanical properties in terms of microhardness, pull-

out and transverse tensile tests, to provide an analysis of the joint

performance, based on microstructural development and the joint’s formation

mechanism;

To generate knowledge from process data, such as process stability, thermal

cycle and microstructural evolution, not only to evaluate the general reliability

of the process, but also to validate further modelling and simulations;

To estimate component performance, based on testing of bolted joints.

By fulfilling the above mentioned objectives, it is intended to propose to the scientific

community a joint formation mechanism based on FHPP of Mg alloys, discussing

both the microstructural and the respective performance changes imposed by the

process. As a result, it is also intended to provide the industry with an alternative

reinforcement process, applicable to a wide range of materials, which may contribute

to the widespread use of Mg and its alloys.

2-5

2 STATE OF THE ART

2.1 GENERAL METAL CHARACTERISTICS

Mg and its alloys, as engineering metals, have always attracted great interest from

the industry. This attraction can be attributed not only to its high specific strength,

good machinability, excellent castability and suitability for high pressure die casting,

but also to its good weldability under a controlled atmosphere, high damping

properties, the possibility of recycling and relatively good corrosion resistance with

high purity alloys [1,2,11]. However, poor cold workabilities, low corrosion resistance,

high solidification shrinkage (in casting alloys), low toughness and high notch

sensitivity as well as their high thermal expansion coefficients are often used as an

argument against the utilisation of Mg alloys. Nevertheless, attempts have been

made not only to improve the characteristic profile of Mg alloys, by employing

different alloying elements, but also to develop new manufacturing/engineering

techniques, which would definitely assure its insertion into the market.

Mg alloys are available in different forms, including cast and wrought. Die casting

provides a good balance between the quality and cost of Mg products and are in

most cases incorporated into vehicles in the form of complex structures with few final

fabrication steps. Wrought products are still expensive, exhibit poor cold workability

at room temperature, due to the limited number of glide systems and the formation of

twins, although some wrought alloys containing zinc or lithium have demonstrated a

significant increase in both ductility and mechanical properties [12]. However, since

Mg is envisaged for use in parts with high safety concerns, there has been a

noticeable increase in interest in wrought alloys in the industry, where casting may

not work as well or needs to be joined [13]. Furthermore, Mg alloys have a closely

packed hexagonal crystalline structure that makes its formability very difficult.

However, at slightly elevated temperatures (200 °C to 300 °C) these materials can be

easily formed and severely worked. The most popular physical attribute of Mg and its

alloys is certainly its light weight and its high strength-to-weight ratios. Its weight is

significantly lower when compared with Al and steels. Additionally, Mg has a

relatively low heat of fusion and a high coefficient of thermal expansion. Such

properties aid in casting, but the distortion during welding can be high [14].

2-6

Various mechanical properties of Mg determine how it will perform in particular

applications. Table 2.1 summarises its physical characteristics compared with other

common structural materials, and Table 2.2 lists its mechanical properties. In the

latter, the competitiveness of Mg and its alloys in comparison with Al and steel

products is clearly demonstrated when the specific properties are considered.

Table 2.1: Physical properties of common structural materials and its alloys [15,16].

Alloy Density [Kg/m3]

Melting Point [ºC]

CTE [μm/m-ºC]

Thermal Conductivity

[W/m-K]

Electrical Resistivity [x10-9m]

Elastic Modulus

[GPa]

Mg 1740 649 26.1 159 43.9 44

Al 2700 660 24 210 26.5 68

Iron 7870 1535 12.2 76.2 96.1 200

Copper 8960 1083 16.4 385 16.8 110

AZ91D-T6 1810 Min. 421 26 72.7 129 44.8

AE42-F 1800 594 - 626 26 --- --- 45

AZ31 1770 605 - 630 26 96 92 45

Al2024-T6 2780 502 - 638 23.2 151 34.9 72.4

Al6082-T6 2700 555 24 180 --- 70

AISI1040(1) 7845 1520 12.5 51.5 221 200

(1) Annealed

In general, Mg alloys tend to undergo a severe reduction in strength at elevated

temperatures. However, alloys containing relatively expensive elements have shown

superior performance at higher service temperatures [17-20] and are therefore

designated to applications where temperature should be considered. Examples of

grades belonging to this group of high temperature resistant alloys are the sand or

permanent mould casting alloys containing zirconium, silver (QE22), scandium

(MgScMn), yttrium (WE43 and WE54), thorium (HZ32, ZH62) and even heavy

elements like gadolinium and dysprosium. These alloys require heat treatment to the

T6 condition. However, some high pressure die casting alloys containing Ca and a

combination of other rare earth mixtures (AE42, AE44, AS+RE) are also suitable to

high temperature applications, and in these cases a further heat treatment is not

required or even not possible. Ductility of Mg alloys is comparable with that of Al and

substantially lower than that of steel at room temperature, but at elevated

temperatures its ductility increases significantly. On the other hand damping capacity

is one of the great attributes of Mg alloys in automotive applications.

2-7

Table 2.2: Mechanical properties of common structural materials and its alloys [15,16].

Elastic Modulus Yield Strength UTS

Alloy Density

[Kg/m3] E [GPa]

Espec.(1)

[km] Rp [MPa]

Rpspec.(1)

[km] Rm [MPa]

Rmspec.(1)

[km]

Mg 1740 44 2.58 x 106 21 1.23 x 103 90 5.27 x 103

Al 2700 68 2.57 x 106 98 3.70 x 103 118 4.46 x 103

Iron 7870 200 2.59 x 106 130 1.68 x 103 262 3.39 x 103

Copper 8960 115 1.31 x 106 62 0.70 x 103 170 1.93 x 103

AZ91D-T6 (chill cast)

1800 45 2.55 x 106 110 6.23 x 103 160 9.06 x 103

AZ91D-T6 (die cast)

1800 45 2.55 x 106 160 9.06 x 103 230 13.0 x 103

AE42-F 1800 45 2.55 x 106 135 7.65 x 103 240 13.6 x 103

AlSi8Cu3 (chill cast)

2700 70 2.64 x 106 160 6.04 x 103 220 8.31 x 103

AlSi8Cu3 (die cast)

2700 70 2.64 x 106 240 9.06 x 103 310 11.7 x 103

AZ31 1800 45 2.55 x 106 155 8.78 x 103 240 13.6 x 103

Al2024-O 2780 72.4 2.65 x 106 75.8 2.78 x 103 186 6.82 x 103

Al2024-T6 2780 72:4 2.65 x 106 345 12.7 x 103 427 15.7 x 103

Al6082-T6 2700 72.4 2.73 x 106 255 9.63 x 103 300 11.3 x 103

AISI1020(2) 7870 200 2.59 x 106 295 3.82 x 103 395 5.12 x 103

AISI1040(2) 7845 200 2.60 x 106 350 4.55 x 103 515 6.69 x 103

(1) Specific properties are related to density and the gravitational constant (2) Annealed (3) Green and yellow highlighted fields are defined for cast and wrought alloys respectively

2.2 NOMENCLATURE AND ALLOYING ELEMENTS

Similar to other materials, pure Mg is rarely utilised in industrial applications. Usually,

it is applied in combination with several others elements, which improve or give to an

Mg-based product the desired property.

Although it is known worldwide and extensively reported in the literature, no

international system for designating magnesium alloys exists. However, there has

been a trend towards adopting the naming method used by the American Society for

Testing and Materials (ASTM) for Mg alloys. Since the ASTM nomenclature is well-

known, not much effort is going to be put in the present work to explain it. Basically,

each alloy is assigned with letters indicating the main alloy elements, followed by

2-8

numbers that indicate their respective percentages by weight. The final letters

indicates the stage of development of the alloy (A, B, C, etc.) and the heat treatment.

This final heat treatment code has to be separated from the rest of the designation by

a hyphen. Those codes, as well as ASTM standardisation for the nomenclature of Mg

alloys, are listed in Table 2.3. Figure 2.1 illustrates a schematic of the most used

grades with their most common alloying elements. The influence of each element on

the final properties of a grade is well described in the literature and so will not be

discussed within this review [2,8,14].

Table 2.3: ASTM codes for Mg’s alloying and heat treatment designations [14].

ASTM CODES HEAT TREATMENT DESIGNATIONS

AL(1) AE(2) Designation Heat-Treatment

A aluminium F As fabricated.

B bismuth

C copper O

Annealed, recrystallised

(wrought alloys only).

D cadmium H3 Strain hardened and then stabilised.

E rare earths H10 / H11 Slightly strain hardened.

F iron H23 / H24 / H26 Strain hardened and partially annealed.

H thorium

J strontium T

Thermally treated to produce stable tempers.

K zirconium W Solution heat-treated (unstable temper).

L lithium T1 Cooled and naturally aged.

M manganese T2 Annealed (cast products only)

N nickel T3 Solution heat-treated and then cold worked.

P lead T4 Solution heat-treated.

Q silver T5 Cooled and artificially aged.

R chromium T6 Solution heat-treated and artificially aged.

S silicon T7 Solution heat-treated and stabilised.

T tin

W yttrium T8

Solution heat-treated, cold worked and artificially aged.

X calcium

Y antimony T9

Solution heat-treated, artificially aged and cold-worked.

Z zinc T10 Cooled, artificially aged and cold-worked.

(1) Abbreviation Letter (2) Alloying Element

2-9

Figure 2.1: Most common Mg alloys with codes according to their ASTM classification.

2.3 GENERALITIES, PHASES AND STRUCTURES OF Mg AND ITS ALLOYS

2.3.1 Magnesium-Aluminium-Zinc (AZ) - Casting Alloys

Mg-Al alloys are among the most widely used industrial die casting applications.

Such alloys present a good combination of castability, mechanical strength, ductility

and excellent corrosion properties. Additions of Al improve mechanical strength and

hardness, with the strengthening effect based on a solid-solution formation and an

Mg17Al12 phase. As well as improving the mechanical properties, Al significantly

enhances the castability of Mg (eutectic system TE=437 °C). However, at high Al

contents, an incoherent, interdendritic Mg17Al12 grain boundary phase is formed

without the formation of Guinier-Preston zones or an intermediate metastable phase,

which causes a rapid degradation in mechanical properties at temperatures beyond

120 °C [21-23]. Further, Mg-Al alloys present low heat and creep resistance with low

ductility at room temperature. As reported on the literature [2,24], the presence of Al

also contributes significantly to the formation of microporosity. The level of porosity

increases with Al content up to 11 wt%, where a peak in porosity formation is

observed, and decreases from there to the eutectic composition. In addition, it has

been reported that the pore size increases with increasing Al content and porosity

changes from a more distributed state to a concentrated small region.

2-10

Mg-Al alloys are often alloyed with Zn, to improve fluidity and room temperature

strength by reducing the solid solubility of Al in Mg. This reduction increases in turn

the amount of precipitated phase formed upon ageing, causing a moderate increase

in strength [8,22]. It also assists in overcoming the harmful, corrosive effects of iron and

nickel. By adding up to 3 % zinc, shrinkage can be compensated for and tensile

strength increased. Additions of Zn in concentrations above 1.0 % – 1.5 % in Mg

alloys containing 7 % to 10 % Al leads to a hot cracking effect [25]. Compared with

phases present in binary Mg-Al alloys, no new phases occur in ternary Mg-Al-Zn

alloys, if the Al to zinc ratio is greater than 3:1 and at no stage does AZ91 contain a

ternary compound. Zinc can also change the eutectic morphology of a single casting

from a completely granular to a fully divorced formation [25].

Manganese does not have much effect on tensile strength, but increases the yield

strength slightly. Its most important function is in improving the saltwater resistance of

Mg-Al and Mg-Al-Zn alloys by removing iron and other heavy metal elements into

relatively harmless intermetallic compounds, some of which separate out during

melting [14].

The microstructure and phase transformation taking place during solidification of Mg-

Al-Zn-(Mn) shows, according to Figure 2.2, the following structural constituents:

hypoeutectic Mg-Al solid solution, eutectic Mg-Al supersaturated solid solution and β-

phase (Mg17Al12). Such phases result from non-equilibrium solidification conditions

that determine the actual solidification process of real ingots and die castings. Under

equilibrium, Mg-Al-Zn and certain Mg-Al-Mn alloys should solidify with the formation

of only hypoeutectic Mg-Al solid solution. However, under non-equilibrium

solidification conditions, the slow diffusion of Al leads to a coring effect in

hypoeutectic Mg-Al solid solution and to the formation of a eutectic β-phase and an

Al-rich eutectic Mg-Al solid solution. In addition to the eutectic precipitates,

discontinuous precipitation can also take place in cast ingots, due to their slow

cooling rate in solid state.

2-11

Figure 2.2: Al-Mg phase diagram [14].

2.3.2 Magnesium-Aluminium-Rare Earths (AE) - Casting Alloys

Due to the fact that Si leads to a deterioration of corrosion resistance and limits

improvement of creep resistance in AS (Mg-Al-Mn-Si) alloys, a new Mg-Al-RE based

variant was developed, which would enhance the creep properties of Mg alloys.

These special alloying elements were added (RE stands for rare earth mixtures) and

a new alloy designated AE42 was developed [8,26].

This alloy was developed from a non-aluminium Mg chemistry in which rare earths,

under die-casting conditions, were shown to increase creep resistance by forming

fine Mg9RE second phase particles along the grain boundaries and some RE-

containing precipitates. This interdendritic phase has been reported to be Al11RE3.

Due to the formation of this phase, the formation of β-phase is suppressed and the

corresponding creep properties of the AE42 alloy are largely improved. Rare Earth

elements presented in commercially available AE alloys typically consist of

Ce (>45 %), La (20–30 %), Ne (10–20 %) and Pr (4–10 %). Neither in “as cast” alloys

nor in heat-treated specimens has the formation of Mg-Al-RE phases been reported [26]. Additionally, both Al-RE-Mn and other Al-RE based phases (Al11RE3 and Al2RE)

are also described in the literature as composing the microstructure of die cast AE42

and of the recently developed AE44 [8,14,27-29].

2-12

2.3.3 Magnesium-Aluminium-Calcium-Strontium (AXJ) - Casting Alloys

The addition of Ca to Mg-Al alloys has a positive effect on grain refining and aids

creep resistance. However, Ca can lead the tool to sticking during casting, degrading

corrosion resistance, castability, ductility and impact strength as well as increasing

the tendency to hot cracking. Since calcium is a relatively inexpensive element and

has an atomic weight, which is half of that of strontium and less than one-third of that

of RE elements, it means that alloying with Ca results in a maximum increase of the

above properties in terms of cost [2,8,14].

With the object of developing a new creep resistant and cost effective die casting Mg

alloy, DSM, Volkswagen and the Technical University of Clausthal established a

research program, resulting in a series of new alloys. MRI153M and MRI230,

developed within this cooperation [30,31], contain around 1 % to 2 % Ca and up to

0.4 % Sr additions. Furthermore the alloy presents corrosion resistance and

mechanical properties at room temperature, similar or better than those of AZ91D

and other commercial alloys. Its elevated temperature strength and creep resistance

are also significantly superior to those of commercial alloys, allowing its use at

temperatures up to 190 °C. MRI alloys can be considered a further development of

the AM alloy with additions of calcium and strontium. Ca produces an effective grain

refinement and increases the creep resistance, as mentioned above, mainly due to

the formation of many fine stable Mg32(Al,Zn,Ca,Sr)49 precipitates, among others. It

has also been reported in the literature that MRI precipitates belong to the Mg-Al-Zn-

Ca system and are expected to be predominantly Al2(Ca,Zn), which are more filigree

than those of AZ91D [32]. Further, the addition of Sr generally causes an improvement

of the mechanical properties and contributes to increases in creep resistance, since

the free Al forms Al4Sr instead of the prejudicial Mg17Al12 at the grain boundaries [33].

2.3.4 Combined Properties

Summarising the results found in the literature, one can conclude that it is very

difficult to develop a new alloy that exhibits an adequate combination of castability,

creep performance, corrosion resistance, ambient strength and affordable cost in

order to meet the requirements, which are necessary to make the use of Mg

widespread throughout the industry. Although AZ91 is the most widely used Mg die-

casting alloy, presents excellent properties at room temperature and its costs seem

2-13

very reasonable, the dissemination of such material is restricted, under certain

circumstances, mainly due to its inferior high temperature resistance. Conversely,

Mg-Al-RE and Mg-Al-Ca-Sr maintain their high strength levels at higher

temperatures. However, the field of application for such alloys is somewhat limited,

principally due to their low castability and sometimes to the high costs involved in the

fabrication of complex components. In Table 2.4, where mechanical properties

versus projected costs for an application at 150 °C are considered, points were

attributed to the different properties, assigning castability and creep resistance the

highest importance. As a second stage, the products of each individual score by its

weight were summed to calculate the total score for a given alloy. Correspondingly,

Figure 2.3 draws a comparison between the most used alloys and their probable

substitutes.

Table 2.4: Combined properties of Mg-alloys for applications at 150 °C [8].

Cast

ability Creep

Resistance

Yield Strength at 150°C

Fatigue Strength

Ductility Corrosion Resistance

Total Score

% of maximum

score

Alloy Weight 10 10 7 7 3 7

Score 9 8 8 10 5 10 MRI153M

WxS(1) 90 80 56 70 15 70 381 100%

Score 6 10 10 9 3 8 MRI230D(2)

WxS 60 100 70 63 9 56 358 94%

Score 5 10 10 8 3 8 AX52J

WxS 50 100 70 56 9 56 341 90%

Score 4 10 9 8 3 9 ACM522

WxS 40 100 63 56 9 63 331 87%

Score 6 9 6 6 5 8 AJ52X

WxS 60 90 42 42 15 56 305 80%

Score 9 7 3 5 10 8 AS21X

WxS 90 70 21 35 30 56 302 79%

Score 7 8 4 5 9 8 AE42(2)

WxS 70 80 28 35 27 56 296 78%

Score 10 1 5 8 5 9 AZ91D(2)

WxS 100 10 35 56 15 63 279 73%

(1) W x S = Weight x Score (2) Highlighted rows are alloys considered within this work.

2-14

Figure 2.3: Combined properties rating vs. projected cost of new Mg alloys designated for applications

at 150 °C [8].

2.4 BOLTED MAGNESIUM JOINTS

As briefly mentioned in the introduction, intensive lightweight design and the use of

modern lightweight metals in the automotive industry are increasing the demand for

new development of fasteners and the design of bolted joints. For many years, Al

and steel components have been joined with high-tensile carbon steel fasteners

without further problems. However, various challenges are experienced with the

combination of both Al and high-tensile steel fasteners with Mg components.

According to the literature, the thermal cycles imposed by operating conditions can

lead to significant loss of the preload force in bolted joints, due to relaxation. In some

cases, losses of up to 95 % are reported in Mg AZ91D-T6 gearboxes fastened with

steel bolts (see Figure 2.4) [9]. Loss of clamp is caused by relaxation of the joint due

to the creep behaviour of Mg itself and the significant differences in the thermal

expansion coefficients between the bolt and clamped material. Figure 2.5 shows the

most used configuration, where the bolt is screwed into an extant thread directly into

the member to be joined, which could be an Mg or Al component [34,35]. Some other

variations involve thread grooving and through-all screws with nuts [36] are very often

investigated and reported on the literature. Additionally, many investigations have

been dedicated to the simulation of bolted Mg joints [37,38].

2-15

Figure 2.4: Load retained as a function of material

combination and operating time [9].

Figure 2.5: Schematic of the most used Mg

bolted joint [34,35].

Operating temperatures on modern powertrain units (e.g., engine and transmission)

are in the range of 150 °C. According to Westphal et al. [34] Mg components joined by

a steel fastener of a standard geometry (M8 flange bolt) experiments an additional

load of 7 kN for a temperature difference of 130 °C (see Figure 2.6). This

corresponds to 50 % of the assembly clamp load for this joint, which means that

significant effects of relaxation due to creep are already observed during the increase

in temperature. Experiments carried out by Westphal [35] using an Mg sample

tightened against a threaded Al specimen of AlSi9Cu3 have shown a clearly higher

remaining load for joints clamped with Al fasteners. Additionally, after 200 hours,

clamp load change was negligible for Al, whereas for steel, losses after 200 hours

were over 90 % and still declining (see Figure 2.7). The tendency to have higher

residual clamping loads with Al fasteners seems to find consent between

investigations [9,35]. According to the authors [9,35], Al fasteners are far more desirable

on a cyclically loaded Mg component joint versus a similar steel fastener, due to the

difference in stiffness ratio of the joint as well as a shorter full thread engagement

(1.2 to 1.5D for Al and 2.5 to 3.0D for steel fasteners, where D is the diameter of the

fastener). A fastener with a large bearing area to accommodate the compressive

yield strength of Mg components would also be helpful, to significantly reduce clamp

load losses.

2-16

Figure 2.6: Theoretical additional clamping load

due to different thermal expansion

coefficients [34].

Figure 2.7: Loss of pre-load force as a function of

time and temperature [34].

During bolt-joint tightening in a bolt-nut-washer experiment, the end of the bolt is

assumed to experience uniform displacement when the bolt is in tension. The bolt

head or washer applies pressure on the Mg joint surface through surface contact and

the joint is under compression. Finite Element Analyses (FEA) show that stresses in

Mg alloy bolt joints are highly non-uniform and the magnitude of stresses is

sometimes close to yield or even above the yield strength. In a bolted joint loaded up

to 72 kN, the stress in the larger part of the sample ranges from 20 MPa to 100 MPa

(compressive stresses). The areas submitted to higher stresses close to contact

corners experiments values up to 110 MPa, which in this case means that plastic

deformation can occur in Mg alloy components under such loading conditions.

On the other hand studies have been directed to foresee the relaxation behaviour in

Mg-sleeve/thread material with different bolt materials [39]. In these cases

compressive stresses imposed by the bolt acting on the Mg-thread experiments show

values near to the Mg yield strength, which is aggravated by high temperatures.

According to the FEA presented in this study [37], the bolt material plays the most

important role on the pretension losses in Mg bolted joints, being the clamping length

the variable with the largest impact followed by the bolt head diameter (bearing area).

2-17

Since creep is the most important mechanism that influences the resistance of

automotive Mg-bolted joints, many authors have investigated high temperature stable

alloys to fulfil this gap, i.e. alloys which have a larger operational range in terms of

temperature [40]. In this context, AJ, AS, ACM, MEZ, AE, WE, Ca-based (MRI) are

examples of suitable alloys, which are being investigated to be used in high

temperature applications [41-46]. Figure 2.8 summarises the tendency observed in

most of the reports, where Mg-Al-Zn-Ca-Sr (MRI) alloys presented the highest creep

resistance, and therefore the highest fraction of remaining load in bolt-load retention

tests, followed by Mg-Al-RE (AE) and Mg-Al-Si (AS) alloys. Mg AZ91 clearly shows

its high temperature limitations and appears as one of the most unsuitable materials

for such applications. Performing a bolt-nut-washer bolt-load retention experiment

Okechukwu [40] observed the alloy MRI230D as the one with the best bolt load

retention properties. In his experiments at 150 °C, MRI230D retained about 32 % of

its initial load, whereas MRI153 and AM50 retained 27 % and 11 % of their initial

loads respectively. Under these conditions, AZ91D experienced permanent

deformation.

Figure 2.8: Creep curves from tensile creep tests of new Mg-Al-Ca-Sr alloys in comparison with

commercial Mg and Al alloys at 150 °C [8].

2-18

2.5 FRICTION WELDING – A LITERATURE REVIEW

As reported by Crossland [47], the first patent related to friction welding was granted to

Bevington in 1891 in the USA. According to his idea, heat would be generated

between a fixed workpiece and a rotating high-speed steel die. In a further stage the

workpiece would soften so that it could be easily extruded through the die. The next

development occurred in Germany during the 1920s, culminating in a German patent

in 1929. These efforts were followed by research in England, which resulted in a

series of patents issued between 1941 and 1944. In September, 1941, Klopstock and

Neelands [48] patented a friction welding process for seam welding. The heat

generated between the rotating welding rod and the members being seam welded

would cause the parent metal to become much more plastic, and induce melting of

the rod. Again, the process apparently did not find wide acceptance. In spite of it

being used for the first time in Germany, to weld thermoplastic piping during the

Second World War, the real credit for the introduction of friction welding on a large

scale is accorded to a Russian machinist A. I. Chudikov, who suggested the process

and patented it in 1956. In 1957 and 1958, many research projects were carried out

in the Soviet Union and other nations, such as Czechoslovakia and China, to develop

and industrialise the process. During this time the first friction welding equipment

were developed by VNIIESO (Scientific Research Institute for Welding Equipment) in

the Soviet Union [47,49,50]. Although friction welding has been around for many years,

the technology and its applications have broadened recently [51].

Friction related processes are versatile enough to join a wide range of part shapes,

materials, and weld sizes. Therefore the technique has become widespread

throughout the world and typical current applications include aircraft and aerospace

components, cutting tools, agricultural machinery, automotive parts, waste canisters,

military equipment and spindle blanks. Furthermore, due to its versatility, the process

has been used to join bimetallic materials in a wide range of applications [52,53]. Figure

2.9 illustrates the actual importance of this subject, by means of the steeply

increasing number of reports and research activities involving the topic of friction

welding during recent years. In Europe, and particularly at GKSS-Forschungs-

zentrum GmbH, Germany, various applied and fundamental R&D projects with

extensive experience on optimisation and characterisation of microstructures of

2-19

friction welded joints have been conducted. The increasing number of publications

highlights the importance of the topic “Friction Welding” and indicates that this

subject is an actual issue and therefore needs further investigation.

Figure 2.9: Number of ISI publications involving friction welding.

Friction welding is a group of solid-state welding processes, which use the heat

generated through relative motion between two frictional interfaces to produce a joint.

This method depends on the direct conversion of mechanical energy into thermal

energy to form the weld. The fixed work piece is held stationary and is brought into

contact with a rotating work piece, normally under a constant welding pressure. After

the rotation stops, an increase in the pressure (upsetting force) can be applied to

improve the mechanical properties [54-56].

Figure 2.10 shows an illustration of the typical friction welding process: In the first

stage Figure 2.10(a) the rotating work piece is brought into movement while the fixed

is held stationary. In Figure 2.10(b) both work pieces are brought in contact and an

axial force is applied to begin the upsetting process (Figure 2.10(c)). In Figure

2.10(d), after the set upsetting is achieved, the rotation stops and the axial force is

increased. The upsetting process continues, since the forging force is applied for

some seconds. As reported by Elmer [55], during the last stage, atomic diffusion

occurs while the interfaces are in contact, permitting a metallurgical bond to form

between both materials, thereby consolidating the joint.

2-20

(a) (b)

(c) (d)

Figure 2.10: Schematic of basic rotational friction welding.

Friction welding involves heat generation through friction and abrasion, heat

dissipation, plastic deformation, and atomic interdiffusion. The interdependency

among these factors leads to complications when trying to develop predictive models

of the friction welding process. The knowledge available is mostly based on empirical

studies performed on a wide variety of materials. The quality of a weld is influenced

by many qualitative factors (bulk material, surface condition, presence of surface

films, etc.) as well as quantitative factors such as the rotational speed, welding

pressure, upsetting/welding time and magnitude/duration of the forging force [50,56,57].

These variables should be controlled as accurately as possible, since they indirectly

govern the mechanisms involved at the different temperature levels and therefore on

the joint’s formation [58,59].

2.5.1 Relevant Process Parameters

According to the literature [51,58,60-63], there are a number of important process

parameters (related directly to the process) and variables (related to the material) in

friction welding. Some of them are listed on Table 2.5. Although all the parameters

are relevant and in some way influence the final quality of the weld, the first four are

the most important and so will be considered in more details. Also, their influence on

the joint quality and process performance will be discussed.

2-21

Table 2.5: Most important parameters in friction welding.

Process Parameters Process Variables

1) Relative speed of the faying surfaces 1) Temperature of the friction surfaces

2) Normal force on the contact area 2) Nature of the material

3) Heating time 3) Presence of surface films

4) Upsetting and upsetting rate 4) Rigidity and elasticity of the friction surfaces

5) Time required to stop the spindle

6) Duration and magnitude of the forging force

Usually, welding parameters are developed and optimised through a series of trials

for a given application. Dennin [64] presented the first attempt to establish welding

parameters for new applications by numerical calculations without performing

welding trials. The reduction of the of welding experiments needed prior to

production is still a matter of research today [65].

2.5.1.1 Relative Speed of the Faying Surfaces

The relative rotational speed is the least sensitive process parameter. In practice, it

may vary over a rather wide range without influencing the quality of the weld

connection. The tendency to increase the rotational speed during friction welding to

obtain better mechanical properties is erroneous, and there are certain optimum

speeds for each individual material combination and application [50,58].

From the standpoint of an improvement of the quality of the welded connection

(particularly for welding of metals sensitive to overheating), it is desirable to use

relatively low rotational speeds. The process efficiency is improved as a result of

reduced heat losses, which reduces the amount of energy used for welding. This,

however, leads to a necessary increase in the power of the installation and, what is

more important, to a much longer initial phase. The duration of this phase (and for

low rotation speeds, throughout the entire welding process) depends to a large

degree on the initial condition of the friction surfaces [50,64,66]. Keeping all the other

parameters constant, Bethlehem [66] reported that when the welds are made with

machined surfaces, the time necessary to complete the initial phase (when the point

of maximum moment is achieved) is significantly shorter. On the other hand the

maximum moment is achieved after a longer time if the surfaces are simply sawed.

Other investigations observed a higher moment peak as the rotational speed

2-22

decreases [64,66,67]. Additionally, Sergin and Sababtsev [68] concluded that the

difference in the distribution of temperatures along the length of steel bars at various

speeds of rotation manifests itself mainly in the first, non-steady stage of the process.

In this stage, at higher rates of rotation the metal along the length of the bar is heated

to a greater depth, and faster, than at low rates.

At high rotational speeds, deep tearing at the friction surfaces is replaced by a

polishing action [50,69]. As a result, to achieve the conditions for plasticisation at the

faying surfaces, longer heating times are required. Longer heating times allow the

propagation of thermal energy along the axial direction of the workpieces, and as a

consequence a greater volume of material is heated. Therefore, in the case of mild

steels, higher rotational speeds lead to lower cooling rates, wider heat affected zones

(HAZ) and hence lower hardness at the vicinity of the bonding line (BL). On the other

hand, lower speed of rotation produces a thinner HAZ, with a profile that is noticeably

much more severely pinched, in comparison with similar welds produced at higher

rotational speeds [60,61,70].

2.5.1.2 Normal Force (Axial Pressure)

Although this variable may vary in broad ranges in the heating and forging stage, it

influences the temperature gradient in the weld zone, the required drive power, and

the axial shortening [64,79]. The axial pressure must be high enough to hold the faying

surfaces in intimate contact, to keep detrimental substances out of the welding zone

and avoid oxidation. However, it should be noted that higher pressure causes local

heating to high temperatures and rapid axial shortening (high upsetting rate), which

might become uncontrollable. Hofmann and Thome [71] observed an increase in the

temperature with pressure in welds performed in mild steel. For a given upsetting,

some studies have shown that the time from the initial contact of the surfaces to the

end of the weld process decreases as pressure increases. With higher pressures, the

consumption of material is faster (higher upsetting rate), generating welds with

shorter welding cycles [60,72,73].

2-23

Following the braking/stopping phase (phase where the rotational speed of the

spindle goes from its maximum to zero), two different methods of varying the axial

pressure are commonly used: maintaining the pressure or increasing it. For the latter

a force called forging force is applied, which according to the American Welding

Society (AWS) [58] improves the joint quality significantly. In general, the forging force

can be applied while the spindle is decelerating during the stopping phase or after

the spindle has stopped rotating at the end of the stopping phase. In both cases a

terminal torque, influenced mainly by the axial pressure, by the rotational speed if the

deceleration rate is high and by the deceleration rate itself, will be generated. As

already mentioned, the second torque peak cannot be eliminated by reducing or

removing the normal pressure before the end of the stopping phase [67,70].

The axial pressure also influences the width and character of the HAZ. Its overall

shape changes from an almost parallel sided boundary at low pressures to a more

‘’pinched’’ or double cone profile at the centre of the stud at higher pressures. In the

second case, the heat liberated is totally used in plasticising the material and, thus,

does not propagate in the axial direction. According to Ellis [60], welds made at higher

pressures show a narrower region, where the hardness values are lower than the

parent material. Figure 2.11 presents the tendency to increase the upsetting rate,

thereby decreasing the welding time, with lower rotational speeds. Additionally a

higher upsetting rate with welding pressure is also evidenced for mild steel.

Figure 2.11: Relationship between welding pressure, speed and upsetting rate [60].

2-24

2.5.1.3 Heating Time (HT)

Vill [50] suggested that it would better to select the duration of the heating period

rather than the upset as the third basic parameter of the friction welding process.

Other authors [68,74,75] suggest the use of temperature in the joint as the third basic

process controllable parameter. The first suggestion is based on the fact that for a

given heat power of the process (speed and pressure are given) the duration fully

determines the energy used in the welding operation. Thus for a given pair of

materials and consequently to a certain extent, the HT determines the temperature

conditions of the process, bearing in mind that the plastic deformation of the

specimen is derived from the temperature conditions. However, the use of

temperature is prompted by the fact that the energy spent in friction welding is used

to create a suitable thermal profile in the interfacial region, which results in upsetting

and bonding.

The HT is defined as the period from initial contact of the faying surfaces to the end

of the braking phase. It is significantly influenced by the axial pressure and by the

rotational speed. It is reduced as pressure is increased and increased with the

rotational speed. For a given pressure, the HT increases with increasing rotational

speed [50,58,60,61,76] and is normally controlled in two ways: the first is with a suitable

timing device that stops rotation at the end of a pre-set time and the second is a

special setting to stop rotation after a predetermined axial shortening (upsetting) is

achieved. HT is also important, especially for low upsetting rate welds, because it not

only defines the microstructure at the interface, but also affects the depth of heating

in the workpieces by conduction and therefore the width of the HAZ. On the other

hand the cooling rate is affected by the surrounding media and the welding time

combined with the total amount of flash formed around the weld interface. If the flash

has a large mass, then the heat stored within the flash will be conducted back into

the weld thereby reducing the cooling rate of the metal within the HAZ. Short heating

times naturally result in a higher heat generation rate. This also involves a relatively

higher upsetting rate, forming a large flash where most of the heat is stored as

sensible energy. By comparison, with longer heating times, not all of the heat

generated at the weld interface is stored in the flash, and much of it has time to be

conducted into the welded parts. Figure 2.12 shows the maximum temperature

2-25

recorded with an infrared camera (see snapshots in Figure 2.12) in the welding

interface of similar rotational AE42 friction welds. This diagram evidences the direct

dependence of peak temperature on the HT.

Figure 2.12: Relationship between welding time and maximum temperature [77].

In order to obtain an improvement in the quality of the welded connection, it is

normally desirable to have short heating times, without flash formation. On the other

hand, when it is desirable to preserve the toughness or not to have high hardness,

for example, it is better to have longer heating times and consequently lower cooling

rates.

2.5.1.4 Upsetting (Axial shortening)/Upsetting Rate

Generally, this variable is not only used to control the welding cycle, but it also has a

significant influence on the joint properties. The applied pressure and speed of

rotation will influence the time needed to reach a pre-set amount of upsetting. The

time from initial contact of the surfaces to the end of upsetting becomes shorter as

the pressure increases and rotational speed decreases. Drews and Schmidt [72]

reported a sharp drop in the upsetting rate as the rotational speed increases (at lower

speeds). On the other hand, at higher speeds (over 3000 rpm) no significant variation

in the upsetting rate was observed with rotational speed. When the upsetting rate is

increased, the total welding time is reduced and hence there is less time available for

grain growth and homogenisation to take place. Figure 2.11 shows the upsetting

rates as a function of the rotational speed and welding pressure.

2-26

Controlling the process by upsetting has disadvantages as well. It has been reported

that large burrs ("collar"), central projections remaining on the surface of the

workpieces, misalignment of the end cross sections (when the friction surfaces are

not perpendicular to the rotation axis) and similar surface defects may have a

negative influence if the process is controlled by upsetting. In these cases, the

welding system considers the wear of these irregularities as the beginning of the

upset process, although the surfaces at this point are not entirely in contact, i.e. are

not yet making full contact. Hence, the control system would stop the weld before the

preset necessary upsetting was achieved [50,78].

Drews and Schmidt [72] have developed a method to calculate the welding

parameters in order to have a defined upsetting. With this method it was possible to

foresee and pre-calculate the parameters to achieve a certain final length of a

component after the welding operation. Generally, Figure 2.13 summarises the

influence of process parameters on friction welding.

Figure 2.13: Parameter influence in Friction Welding.

2.5.2 Process Characterisation – Phases of the Process

The division of the process into different phases is a common way when explaining

the weld cycle and the mechanisms related to the welding process. Various authors

divide the cycle into different numbers of phases. Two [58], three [47,50,79], four [62,64,80-

82] and even five [60] different phases are proposed in the literature to describe the

process. The division in four phases is used by most of the authors and seems to be

the most suitable to be presented in this review.

2-27

2.5.2.1 Phase 1 – Rubbing Phase

During the first part of phase 1 (see Figure 2.14) the initial contact of the faying

surfaces mainly results in a smoothening effect of the surfaces. The remaining phase

1/2 can be briefly characterised by seizing and microbonding processes [47,50,83].

At the beginning of phase 1/1 only a limited amount of the nominal welding surface

actually makes contact. Therefore, the surface pressure reaches extremely high

values locally, promoting plastic deformation and hence, flattening of the faying

surfaces. In addition to these relatively high stresses, the outer regions of the joint

are subjected to the highest relative rotational speed between the faying surfaces. By

rubbing, the surface roughness is partly smoothened by elastic and plastic

deformation and local melting starts in the outer regions of the weld zone [47,50,84]. As

the process progresses, isolated local bonding (microbondings) that eventually takes

place across the contact surface are immediately sheared off. The shearing of those

microbonds induces additional heating in the adjacent material since the energy used

to deform them is liberated in form of heat [50,85]. Neumann and Schober [81] explain

this shearing-off phenomenon by the fact that once the temperature has quickly, but

only locally, reached the melting point or even higher values, the heat generation

decreases substantially. The heat dissipation increases and melting cannot be

maintained, leading the material to plasticisation. When the material has plasticised,

the recrystallisation rate is smaller than the deformation rate, which results in

shearing-off the local microbonding. The mechanism of the energy input and

transformation is also presented by Bethlehem [86]. As the material reaches the

melting point for a short period of time, followed by a sharp temperature drop caused

by high thermal gradients, low transformation temperature products (e.g. martensite)

might be formed in the weld zone. However, Bethlehem [86] reported that even with

longer heating times, no molten film was observed. Kreye [87] reported in some cases

a molten film may be formed locally, although it is not necessary to build the joint.

2-28

The main effect of phase 1/1 is the smoothening of the outer regions of the rubbing

surfaces, which takes place during the first revolutions of the work piece within

phase 1. When the smoothening in the outer regions is almost completed, it starts to

develop towards the rotational centre of the joint (gradual transition from phase 1/1 to

phase 1/2). In the outer regions, almost no heat is produced during this intermediate

phase, because of the reduced frictional contact (the faying surfaces are now widely

smoothened) and the presence of oxide layers generated in phase 1/1. Bethlehem [86] observed that the region between 0.3R and 0.7R, where R is the radius of the

workpiece, is subjected to severe deformation and heating in the beginning of this

second stage. According to his work, the heat losses in the outer regions and the

lower heat generation in the inner part of the welding cause, at the beginning of the

rubbing phase, this localised increase in the temperature. Eichhorn and Schaefer [70]

observed that heating and plasticisation at the beginning of the process occurs more

preferably in the rotating partner, which leads to an uneven HAZ and degree of

deformation.

Figure 2.14: Phases of the process: 1) Rubbing phase; 2) Heating phase; 3) Braking phase and 4)

Bonding phase.

2-29

Phase 1/2 is completed as soon as the friction moment reaches its maximum. This

peak is caused by the removal of the contaminant film on the one hand and by a

significant increase in the number of microbonds distributed along the faying surfaces

on the other. The extraction of the contaminant film allows contact of fresh metal

surfaces with a considerable increase in the coefficient of friction. On the other side,

as a consequence of the increased number of microbonds formed, since a larger

common area is brought into contact, the friction moment necessary to shear such

regions off is also significantly higher. The temperature of the friction surfaces

increases accordingly. Bethlehem [86] observed a considerable influence of the

rotational speed on the maximum frictional moment. However, the axial pressure

causes almost no change in the magnitude of such a parameter. Additionally, it was

also reported the influence of innumerable external variables, such as the

preparation of the faying surfaces, the presence of surface contaminants, oxidation,

grease and oil also affect the magnitude of the maximum friction moment.

2.5.2.2 Phase 2 – Heating Phase

The hydro-extraction effect of surface contamination particles closer to the rotational

centre is obstructed by lower radial forces and restricted material flow conditions.

Therefore, a material concentration on a circular ring area can be observed between

0.5 and 0.7 of the friction surface radius. This concentration causes local heating

forming a thin plasticised film of sheared-off material [62,84,86]. This ring area increases

in size and causes the squeezing of highly plasticised material to the cooler zones of

the friction area. The temperature in this area is comparatively low, because the

radial deformation is quite small due to the low relative speed. In this case, particles

encapsulated in the ring and not re-dissolved into the plasticised material are

transported to the rotational centre of the welding zone. The widening of the ring area

continues until it reaches the rotational centre, forming a circular area, now extending

its diameter slowly towards the outer regions of the weld zone. When the faying

surface is fully plasticised, temperature equilibrium is achieved, which is supported by

a self-balancing effect and causes a decrease in the friction moment. This effect can

be described as follows:

2-30

“As the temperature rises, the metal becomes more plastic, the friction moment

necessary to disrupt the microbonds decreases, due to the lower yield strength of the

materials resulting in a reduced torque, which gives a lower heat generation with a

resultant lower temperature. This in turn softens the material, resulting in less

deformation work and in a local decrease of temperature. Owing to the axial

shortening, fresh material comes to the frictional interface, which causes an increase

in the coefficient of friction leading to an increase in the temperature” [47,50,86,88-90].

Subsequently the temperature rises to a maximum during the first 10 seconds of

frictional contact and remains at a somewhat steady state until the end of the heating

phase. As the process stops, the generation of heat is reduced, causing the joint to

cool down.

In the outer regions of the interface, the material is easily pressed out of the frictional

surfaces, due to the high plasticisation and to low deformation constraints. As the

temperature of the material and the height of the plasticised zone increase, the

resistance to the axial pressure reduces and the material is pressed out of the friction

area forming the flash. This process is maintained as long as the temperature and

the height of the plasticised zone can be kept in balance. As a result more material is

pressed into the flash. These balanced actions result in heat saturation of the friction

area and the corresponding zones. From this point on, an almost constant upsetting

(i.e. shortening of the workpieces) is established.

With increasing thermal saturation of the workpieces, the required energy for

plasticisation reduces, leading to a lower frictional moment. However, the growing

flash increases the frictional surface and hence the frictional moment. The common

opinion is that these phenomena at some point balance each other, resulting in an

almost constant friction moment. Phase 2 is concluded as soon as the process

reaches the limiting control parameter, which means a predetermined temperature

point, a certain amount of upsetting or a pre-set time.

2.5.2.3 Phase 3 – Braking Phase

The braking phase begins with a controlled (when using a continuous drive unit)

decrease of rotational speed (Phase 3/1). Reduced rotational speeds increase the

number of microbondings that simultaneously takes place along the frictional

2-31

surfaces. The shearing of those increased number of microbonds produces

additional heating, which keeps the temperature raising. Additionally, the shear

resistance increases due to the lower deformation rate and hence the friction

moment reaches its second peak. It has been published some literature [91,92]

investigating the deceleration time in friction welding of mild steels. A conventional

friction welding machine was used with the brake pressure varying. In some tests,

the brake was disconnected, allowing the friction at the rubbing surface to stop the

component. It was found that the second peak of torque and the upsetting rate

increases as the brake pressure decreases at the end of the heating phase.

Furthermore, it has been found [70] that this terminal torque cannot be prevented by

reducing or removing the axial force before the end of rotation. Duffin and Crossland [93] have indicated that in continuous drive welding it is not essential to use a brake to

produce welds free of defects. The increase of the frictional moment causes

additional deformation in the adjacent material, which had previously not been

subjected to any deformation. It should be noted that the deformation resistance of

these areas is comparatively low, since they have almost reached the friction surface

temperature (due to thermal saturation in the axial direction) before they are actually

subjected to deformation. At a critical point, the rotation is reduced to such a low level

that the temperature at the faying surfaces can no longer be maintained. Therefore,

the shear resistance of the material increases and torsional deformation takes place

in a larger region along the workpieces. The spindle stops followed by the beginning

of the recrystallisation process, which occurs during the end of this phase resulting in

a microstructure with very fine grain size [62].

2.5.2.4 Phase 4 – Bonding Phase

The mechanism of bonding already begins in the final stage of the heating phase,

although bonding is not homogeneous across the contact surface. A homogeneous

bonding in the whole cross section is achieved by the so-termed forging pressure

within the bonding phase. Although the most authors state that the use of forging

pressure significantly improves the mechanical properties of the weld, some

experiments show that good welds can be obtained without an increase of the force

after rotation stops, which means without any forging force [60].

2-32

The change from the friction to the forging force should take place at the end of the

braking phase. The increased axial force results in an abrupt increase of the

upsetting rate, and consequently of the frictional moment. Oxide layers in the outer

regions of the weld zone produced by insufficient frictional contact are now expelled.

The welding surfaces make closer contact and particularly the materials in the outer

regions are brought together within atomic distances to produce metallic bonding.

After the rotation has stopped, the dynamic softening process continues, i.e. the

deformation is complete, but the diffusion processes still remain. As the material

begins to cool down, static recrystallisation, crystal regeneration and creeping

processes occur. Internal stresses are widely eliminated - a determinant factor for the

mechanical properties of a friction weld [62].

2.6 FRICTION HYDRO PILLAR PROCESSING

Friction Hydro Pillar Processing is, when compared with other conventional friction

techniques, a very recent welding process. Patented by TWI [94], this variation on

friction welding can be described as a drill and fill process, and since its invention has

been investigated in many publications [78,95-108]. Although approximately 15 studies

have been published on this topic, there are basically only two groups investigating

the process worldwide. Furthermore, studies published in various conferences have

fundamentally the same content, which significantly limits the information available

and makes a sound literature review on the topic a difficult task. Bearing that in mind,

the most comprehensive description was given by Thomas and Nicholas [101], while

an illustration can be found in Figures 2.15 and 2.16.

“The FHPP technique involves rotating a consumable rod co-axially in an essentially

circular hole whilst under an applied load, to generate continuously a localised

plasticised layer. The plasticised layer coalesces and comprises a very fine series of

adiabatic, helical rotational shear interfaces, partly spherical in shape. During FHPP

the consumable member is fully plasticised across the bore of the hole and through

the thickness of the work piece. The plasticised material develops at a rate faster than

the axial feed rate of the consumable rod, which means that the frictional rubbing

surface rises along the consumable to form the dynamically recrystallised deposit

material. The plasticised material at the rotational interface is maintained in a

sufficiently viscous condition for hydrostatic forces to be transmitted, both axially and

radially, to the inside of the hole, enabling a metallurgical bond to be achieved.”

2-33

Although filler metal, flux and shielding gas are not required, it has also been shown

in the literature the use of a shielding gas as a method to reduce possible

atmospheric contamination. In this action, the pillar, which in the present work will be

responsible for the local component reinforcement, will undergo significant hot

working when subjected to these severe deformations. Consequently, the formation

of a refined heat-treated microstructure will result in a remarkable change of the

static and dynamic properties of the consumable rod and the parent plate material.

However, such microstructure changes can be modified by thermal treatment later

on, thereby achieving the required properties.

Figure 2.16: FHPP principle.

Figure 2.15: Schematic of FHPP process. (a) Pillar

(b) Frictional Interface

(c) Extruded Material

The developments that lead to the invention of FHPP have been described by

Andrews and Mitchell [109]. This report covers the development of the friction plug

welding process for the repair of offshore structures. The friction taper plug welding

(FTPW) process involves drilling a tapered hole through the full thickness of a plate

at the location of a defect and forcing a rotating tapered plug with similar (but not

necessarily identical) taper angle into the hole. The heat generated from the contact

between the plug and the hole’s wall causes the material to soften and flow, and the

complete conical surface of the tapered plug is welded to the matching hole surface

almost instantaneously [110]. The need to keep the equipment size down, but retain

the possibility to repair thick sections, later led to the invention of FHPP [97]. Figure

2.17 depicts a macrograph of an FTP weld in Al AA6082-T6.

2-34

Cross sections of FTP welds (see

Figure 2.17) showed that the

consumable only plasticises in the

contact area near to the

surface [109,110], while in FHPP the

consumable plasticises across the

whole bore of the hole through the full

thickness of the section [96,105,106] (see

Figure 2.18).

Figure 2.17: Macrograph of a FTP weld [110].

The first comprehensive report on FHPP, including microstructure and mechanical

properties was compiled by Nicholas [97]. The first part of this report covers the

process itself, where a parallel hole configuration is presented, as well as a tapered

hole arrangement. Nicholas [97] found that those “materials which do not exhibit

adequate plastic flow characteristics, often respond much better to a tapered joint

design”. The bonding mechanism is described by metallurgical bonding, owing to a

normal force on the interface between the plasticised material of the stud and the

base material. The formation of a normal force at the sidewalls, although only an

axial force is applied, is explained by hydrostatic behaviour of the plasticised material

inside the hole [96]. Nicholas [97] assumes that a tapered hole configuration will assist

the bond formation due to the additional forces on the tapered surfaces.

Investigations into the influence of the taper angle on the bond quality have not been

reported so far. Furthermore, as the consumable rod undergoes significant hot

working during the welding process, a very refined, hot worked microstructure was

reported. Owing to the superior static and dynamic properties of the extruded

material, FHPP was suggested as a method of enhancing materials as well as for the

production of monolithic materials and MMCs [95].

2-35

In the second part of his report, Nicholas [97] presented microstructural and

mechanical results of a mild steel stud welded in a C-Mn steel substrate. Depending

on the process parameters, either a very refined microstructure across the joint or

some refined regions, which have not undergone layer shearing, could be found.

These characteristics were attributed to the balance between thermal transfer and

material displacement. If the balance results in a heat flow along the stud, a part of

the stud will shear off due to thermal softening. Nevertheless, it was reported that

both weld specimens had good bending and tensile test properties, while the exact

values were not given. In addition, welding trials with aluminium alloys showed poor

bonding quality with parallel hole configurations, while the change to a tapered hole

and stud shape resulted in good bonding properties. This was explained by the extra

welding force generated by the tapered surfaces and the reduced tendency to shear

due to the increasing stud diameter [97].

The first investigations and welding trials with a portable FHPP welding head have

been described by Pauly [98]. In this work it has been successfully demonstrated that

sound FHPP welds could be produced with portable welding equipment. The main

difference in the welding equipment used by Pauly [98] to the one used by Nicholas

and Thomas [94-96,101,103] is the possibility to apply rotational speeds up to 7000 rpm.

Pauly [98] studied a number of different stud and hole configurations with encouraging

results, although still leaving some lack of bonding at the bottom area of the hole at

the end of the investigation. Aiming to further develop the process and assess the

joint properties, Meyer [78] investigated similar welds using API 5L X65 (a HSLA C-Mn

steel) as the base material. In this case, the shear planes mentioned by Nicholas [97]

could be perfectly visualised with nickel tracer rods experiments (see Figure 2.18)

carried out by Meyer [78]. The only difference between both approaches is that

Nicholas [97] used big, stationary workshop machines, limited to low rotational speeds

and high axial pressures, while the system employed in Meyer’s work makes use of

high rotational speeds and low axial pressures.

2-36

The nickel tracer in Figure 2.18 shows

that at the beginning of the weld a

significant amount of stud material is

plasticised and pressed into the gap

between stud and hole. According to the

author [78], in the bottom area the nickel

tracer is moved out of the centre over a

wider area. This happened another three

times along the weld with reduced

amounts of nickel material. Those three

areas depict shear planes. Between the

shear planes the plasticisation of the

nickel is considerably less. This means

that after a certain amount of plasticised

material is produced at the beginning of

the weld, it is pressed up the gap

between stud and hole. Owing to the

frictional heat the stud material is heated

as well. The plasticised material in the

gap cools down in contact with the cold

plate and increases the torque on the

stud. As reported by Nicholas [97], this

results in shearing of a certain amount of

stud material. After this the process

continues on top of the sheared stud

material, creating a new HAZ, which can

be seen in Figure 2.18. Meyer [78] further

suggests that the number of shear

planes and the distances between

depends on the process parameters and

the material properties (i.e. tensile

strength at elevated temperatures).

Figure 2.18: Macrograph of an API 5L X65 weld

with a nickel tracer rod [78].

2-37

The patent on the process was granted to TWI in 1993 [94]. Thomas and Nicholas [95]

presented further advantages of the process in the same year, focusing on its

application in the heavy industry joining or repairing of thick steel structures. It was

stated there are special advantages for FHPP with thick-walled structures, due to the

fact that the amount of weld material needed is comparably low and, as in all friction

welding processes, FHPP is readily mechanised and easy to automate.

The main application areas were indicated as the repair of steel structures in the

construction as well as in the oil & gas industries, where some tests have already

been done in order to use the process commercially. Single FHPP welds and Friction

Stitch Welding (FStitchW - a series of successive FHPP welds) are being developed

for sleeve welds and for hot-tapping saddles onto a large diameter pipeline or a

substrate with gaps varying between zero and four millimetres. Recently,

investigations have confirmed in both single FHPP (see Figure 2.19) and FStitchW

(see Figure 2.20), the feasibility of the process to perform sound welds without a lack

of bonding even in adverse conditions [100,111]. To date, any current investigations or

research programs involving the application of FHPP in lightweight alloys have not

been published.

Figure 2.19: Macrograph of a single

FHPP sleeve weld [100].

Figure 2.20: Macrograph of an FStitch weld conceived for

the repair of large diameter pipelines [111].

The FHPP-related nomenclature is not fully defined so far, with a lot of similar names

used for different process variations. The following list tries to define some of the

terms related to FHPP. These phrases will only be used with these given meanings

within this work.

2-38

Friction Hydro Pillar Processing (FHPP): Friction welding process where a

rotational welding consumable is friction welded in a bore hole. The welding

consumable and the bore hole can be of different geometries and shapes, including a

tapered configuration. The welding consumable plasticises across the whole

diameter and across the whole welding depth.

Friction Tapered Plug Welding (FTPW): Friction welding process where a through-

thickness hole is closed by friction welding a rotational welding consumable to the

sidewall of the hole. The welding consumable is only plasticised near the surfaces

and not across the whole diameter. Usually, a tapered configuration for the

consumables (plug) and hole is used.

2.7 FRICTION WELDING IN MAGNESIUM AND ITS ALLOYS

As presented in the earlier sections, Mg and its alloys play actually an increasing role

mostly in the aeronautic and automotive industry. In order to further extend the

number of applications where such alloys can be employed, the development of new

manufacture procedures and an improvement of the available joining processes are

necessary. Although its potential has already been demonstrated, Friction Welding of

Mg alloys in Europe has been a lesser explored area, with no more than 7

publications [112-118]. However, most of those contributions deal very superficially with

the subject, principally in terms of microstructural changes. Some other reports deal

with dissimilar weldability of those alloys [119] while the rest show developments in the

friction stir welding of Mg in similar and dissimilar configurations [120-124]. As a review,

the main aspects presented in these publications will be discussed, however,

keeping in mind the lack of specialised literature on the subject.

Most of the authors consider the friction weldability of Mg and its alloys as possible,

although some references indicate it as not weldable [125] with direct drive welding

systems (the motor remains attached to the workpiece during the whole welding

cycle). According to Bowles et al [112], it is possible to join AE42 to AZ91 by friction

tapered plug welding without considerable loss of hardness across the bonding line.

On the AZ91 side of the weld, more significant microstructural changes are observed,

since dynamic recrystallisation occurs from the weld line to approximately 500µm on

this side of the weld. On the other hand very few changes in the microstructure are

2-39

evident on the AE42 side. Only a mechanical fragmentation of the intermetallic

particles is observed. Such a process was used to improve local mechanical

properties of an AZ91 cast component by joining AE42 studs. A series of

investigations, involving not only the results reported by Bowles et al [112], but also

many experiments on similar and dissimilar weldability using different friction welding

processes (see Figures 2.21 to 2.24) were considered initial approaches, which have

resulted in this study.

Figure 2.21: Picture of a similar FTP weld using

AZ91D-T6 as base material [126].

Figure 2.22: Macrograph of a dissimilar FHPP

weld using an aluminium AA6082 stud and a

magnesium AZ91D-T6 base plate [126].

Figure 2.23: Macrograph of a dissimilar friction stud weld between a magnesium MRI153M and a

magnesium AZ91D-T6 plate [127].

Figure 2.24: Macrograph of a dissimilar rotational friction weld between an AA6082 and an AZ91D-T6

stud. The presence of a pure aluminium AA1050 interlayer was also investigated [128].

2-40

Ogawa et al. [113] have shown, through several experiments in a wide variety of

friction welding conditions with similar AZ31 welds, that the deformation heat input in

the upset stage and upset loss can be used to evaluate the joint performance in

terms of tensile strength. In this study, the minimum values of heat input and axial

shortening to obtain sound joints are more than around 1500 J/sec and 17 mm

respectively. Friction time, rotational speed and upset pressure must have expressive

values for the welds to lie in the sound joints range. According to the author [113], in

order to obtain an elongation and reduction of area equivalent to those of the base

material, it is recommended to set a fairly high friction pressure and forging force

without any excessively long friction time or high rotational speed.

Similar AZ31 joints were also studied by Kato and Tokisue [114]. According to the

authors, the macrostructures in the vicinity of the weld interface are symmetrical in

relation to the weld interface and joint axis regardless of the applied friction time. The

fibrous structure running parallel to the extrusion direction, as observed in the base

metal, shows a change of structural flow into the flash discharge direction, i.e. the

circumferential direction. It was also reported that this fibrous structure disappears in

the region very near to the weld interface. Concerning mechanical properties, the

hardness is, in the weld interface, equivalent to that of the base material and seems

not to significantly vary along the weld interface. Furthermore, the authors assure

that tensile strength and elongation tend to be improved with an increase in the

friction pressure and heating time.

Pinheiro et al. has demonstrated in a series of studies [73,77,126,127,129] the feasibility of

the process to join Mg-Al-RE alloys. These investigations have demonstrated that

sound welds can be obtained, with a weld region characterised by a thin layer of

dynamically recrystallised grains without any unexpected particles or phases

precipitated at the grain boundaries. Additionally, SEM analyses have indicated a

complete disintegration of the typical lamellar BM structure along the recrystallised

zone, extended partially to the thermo-mechanically affected zone. Fully satisfactory

bonded joints, in terms of tensile strength, were obtained with no joint collapsing at

the welding line.

2-41

Friction weldability of Mg alloys has also been investigated recently by Draugelates

et al. [115,116,118]. According to these experiments, an increasing rotational speed

reduces the upsetting rate up to a value where the speed has no further influence,

i.e. the upsetting rate achieves, after a certain speed, a quasi-stationary value and

cannot be further influenced by changes in the rotation. Similar to the investigations

carried out by Kato e Tokisue [114], Draugelates et al. [118] observed that the

microstructure in magnesium friction welded cast alloys is completely recrystallised

along the welding area, with the grains reoriented to a direction parallel to the

welding interface. Furthermore, in the AZ91 alloy, the Mg17Al12 phase, previously

present in the grain boundaries, is dissolved along the bonding area due to the

higher temperatures achieved during the process. In terms of mechanical properties,

heating times of 20 s lead to the highest performance welds, with respective failures

taking place always in the base material outside the welding area (see Figure 2.25).

Figure 2.25: Tensile strength vs. heating time [118].

3-42

3 EXPERIMENTAL

As presented in Figure 3.1, the experimental strategy adopted in this study consisted

of an initial exploratory development of the FHPP technique. In this initial phase,

different materials and several geometries were investigated using a range of

parameters, set in order to determine the outline of a suitable window in which this

study would be further developed. After this initial stage, three materials were

selected for the main experimental phase, based on the commercial importance

associated with its use within the objectives proposed in this work.

The second experimental phase consisted of welding these materials, divided into

two groups: In the first group, similar welds were produced while varying the welding

pressure, considered the most influential welding parameter. As the effects of the

welding pressure would be investigated by the first group, three material

combinations were welded in a second group, using two upsetting variations with all

other variables kept constant.

In the main experimental phase, the work was divided into four topics: Process,

Microstructural Development, Properties and Performance. For each topic, suitable

tests were proposed and performed. By coupling the single results of these individual

topics, conclusions on the bonding mechanisms, structure/properties relationships

and service behaviour were arrived at. Based on the information gained during the

work, a design catalogue has been outlined, where the resultant properties are

compiled, based on the process welding parameters.

3-43

Figure 3.1: Summarised experimental strategy of the present work.

3-44

3.1 BASE MATERIALS

Three different Mg alloys have been used within this study. For all the experiments

the alloy AZ91D-T6 was used as the base plate. The material was delivered in the

form of cast ingots, heat treated (413 °C/20 hs) and artificially aged (168 °C/16 hs) to

T6 condition according to DIN65582 [130] (ASTM B 661-03) [131] since the mechanical

properties of AZ91-F, as fabricated, is shown to vary across a very wide range.

For welding studs, AZ91D-T6, MRI230D and AE42 cast ingots were used. Further

heat treatment of MRI230D and AE42 ingots (see Figure 3.2) was empirically

estimated, so as not to exert any positive influence on the performance of the alloys.

Therefore, in both cases the material was used in the “as fabricated” condition. A

detailed description of the microstructure and properties of the base materials is

presented in Section 4.1. Cast ingots have a coarse microstructure with consequently

limited mechanical properties. Despite this, the base materials were chosen in this

form, since the changes imposed by the welding process would significantly enhance

their properties.

3.2 STUDS AND HOLES CONFIGURATION

Based on previous studies carried out by Pauly [82] and Meyer [78], tapered studs

(reinforcing material) with a 10° taper angle, 6 mm and 8 mm tip diameter studs were

chosen. Similar and dissimilar configurations involving these two different stud

geometries were tested. Figure 3.3 shows one of the stud configurations (8 mm tip

diameter) used within this study.

Holes with 10° and 20° taper angles and with 6 mm and 8 mm bottom hole diameters

were used. Figure 3.4 illustrates a hole with 20° taper angle and 8 mm bottom

diameter. Lateral openings were drilled in the base plate with 0.5 mm distance to the

borehole at 6 mm and 11 mm from the plate surface. Type K thermocouples with

1.0 mm diameter were then placed in these openings and connected to the

thermocouple measuring system (described in detail in Section 3.4.2).

3-45

Figure 3.2: AE42 and MRI230D Ingots.

(a) (b)

Figure 3.3: In (a) the configuration of the tapered stud and in (b) the stud clamped on the welding head

before the welding.

(a) (b)

Figure 3.4: Hole geometry combinations.

3-46

3.3 WELDING SYSTEM

The welding system used, whose components are briefly introduced below, can be

described as a hydraulic powered friction welding machine, which was designed and

built as a portable equipment for underwater use by Circle Technical Services Ltd. in

Aberdeen. The system is capable of applying 40 kN axial load at speeds up to

8000 rpm (see Figure 3.5) and consists of four major components: Hydraulic Power

Unit (HPU), valve block, welding head and control system. The weld cycle is

electronically controlled, with the functions monitored and displayed on-line on a

monitor. The welding can be classified as a direct drive system in which the friction

welding process is controlled by the upsetting distance, which is the amount of

plasticised material pressed out during the welding cycle (distance that the hydraulic

ram moves out).

HPU: In the configuration used, the HPU’s pump is driven by a hydraulic power pack

with a 50kW electric motor, supplying up to 115 l/min at 315 bar. The maximum oil

pressure is continuously adjustable, while the pump automatically adjusts the flow.

Valve block: The valve block regulates the oil flow and pressure supplied by the

HPU. Electrically driven proportional valves on the valve block control all operations

of the weld head. Besides additional safety valves, three pressure transducers are

used to monitor the mainline pressure (supplied by the HPU) and the two sides of the

piston in the welding head (axial force). Supplementary to this standard configuration,

a number of additional sensors were installed, to allow a comprehensive monitoring

of all system variables. Two pressure transducers are used to measure the pressure

across the hydraulic motor of the welding head.

Welding head: The portable head itself is approximately 600 mm long and 160 mm

in diameter. It consists mainly of a hydraulic fixed-displacement motor on the upper

side and a hydraulic piston inside the lower cylinder. The piston can move the lower

part of the shaft by a nominal 50 mm in the axial direction and applies the pre-set

axial force. Welding studs (consumables) are fixed in the chuck at the lower end of

the shaft. A metal proximity sensor on the shaft inside the housing monitors the

rotational speed, while a linear proximity sensor (LPS) measures the axial movement

inside the piston.

3-47

The weld head is attached to a welding rig and support structures by a bayonet

connection on the housing. Figure 3.5 and Figure 3.6 show respectively the welding

head itself and the complete welding equipment mounted on the rig.

Figure 3.5: Circle Technical Service's HMS 3000

Welding Head.

Figure 3.6: Welding equipment.

Control System: A PC-based control system acquires the sensor information

(rotational speed, axial position and piston pressure) and controls according to the

settings of the welding parameters. This control system regulates the welding

process’s force and rotational speed. The upsetting rate is a result of these

parameters. The pre-set welding parameters (rotational speed and axial pressure)

can be changed according to the actual upsetting a number of times during one weld.

This gives a very flexible tool to respond to changes in the welding conditions during

a single weld (i.e. changes in geometry). Process parameter sets are saved as

“projects” by the control system. They have to be defined before the actual weld

takes place. The welding parameters are displayed in real time during a weld and

stored within the projects.

3-48

3.4 DATA ACQUISITION SYSTEM (DAS)

3.4.1 Data Recording System – DRS/Process Stability

As mentioned in Section 3.3, an additional measurement system was employed for

acquisition and processing of process parameters and complementary data. This

system, denominated the Data Recording System (DRS), was used to measure and

store the welding parameters and additional signals for every weld (rotational speed,

axial position of piston, hydraulic position on both sides of the piston, main line flow,

the oil temperature in the main line, main line hydraulic pressure and hydraulic

pressure to and from the weld motor). The system is based on a standard PC

connected to National Instruments hardware for signal conditioning. The software

was programmed in LabView and gives real-time diagrams from all signals generated

during the welding. Data can be stored and exported into standard data processing

software for further analysis and for complementary evaluation. Owing to the fact that

all sensors are processed by a single unit, including signals generated by the welding

control system, reference across all data (welding parameters and system variables)

is possible. The DRS records data at 100 Hz per channel. Data files were stored for

every weld produced in this study. These files were than evaluated and a precise

analysis of the stability and consequent repeatability of the welding system within the

welded range could be established.

3.4.2 Temperature Measurements

For experiments on thermal phenomena, a special measuring system based on

signals generated by thermocouples was devised, programmed and assembled. This

is a PC-based measuring and data acquisition unit. The PC is connected to National

Instruments SCXI-hardware for signal conditioning. The software was programmed in

LabView and generates real time diagrams during measurement. Data can be stored

and exported for further evaluation and analysis. Up to 32 thermocouples can be

recorded simultaneously, with a maximum of 50 Hz per channel. Temperature

measurements were started simultaneously with the HMS control system and all

recorded at 50 Hz. Calibration curves for the different types of thermocouples were

already implemented in the LabView software package to convert the electromotive

force (mV) into the appropriate temperature (°C).

3-49

As previously presented in Figure 3.4, FHPP welds were performed using 2 type K

thermocouples with 1.0 mm diameter placed 0.5 mm away from the borehole at

6 mm and 16 mm from the top base plate surface and connected to the

thermocouple measuring system. The main objectives of these measurements were

to establish the influence of welding parameters and geometry combinations on the

temperature cycle and therefore on mechanical properties of the welded joints.

3.5 TESTING EQUIPMENT AND PROCEDURES

3.5.1 Welding Procedure

The welding studs and plates were permanently marked, prior to preparation for a

welding trial. Studs and plates were then degreased and cleaned with Acetone to

avoid any influence of machining fluids. The stud was fixed in the weld-head chuck

and aligned with the hole in the base plate. All welding parameters were previously

typed into the control system and could then be loaded by reference number. The

DRS system and the weld control are manually started and the weld is performed

according to the given welding parameters. The rotational speed is started first and

after the required speed is reached, the piston is moved down and the stud then

moves into the hole and makes contact with the cavity bottom. From this point on, the

axial movement is measured as upsetting and is used to control the welding process.

When the full upsetting distance is reached, the rotational speed is stopped and the

axial pressure is increased for a further six seconds. Afterwards, the pressure is

released and the specimen can be taken out of the fixtures.

3.5.2 Metallurgical Characterisation

3.5.2.1 Optical Microscopy

Macro and microstructural characterisation has been conducted on both base

materials and cross-sections of welded specimens. This assessment included

different welding areas following the ASTM E3-95 [132] and ASTM E-340-95 [133]

standards. Welded specimens were cut at the centre of the stud across the base

plate, followed by grinding and polishing with diamond paste down to 0.05 μm using

a Phoenix 4000 automatic polishing machine and based on standard metallographic

techniques. Two different etching solutions were subsequently used for

3-50

characterising the joints; first an acetic glycol based solution, used to assess grain

size, and another picric acid based solution, basically used to reveal details of the

microstructure across the welded samples. Micro and macrographs were made using

a LEICA DM IRM microscope equipped with the software LEICA Q550IW. Chemical

composition was ascertained using inductively coupled plasma source mass

spectrometry (ICP-MS), which is a very sensitive analytical method for rapid multi-

element determination. The quantisation limit was in the worst cases 0.01 % for

silicon followed by iron with 0.005 % inaccuracy.

3.5.2.2 Grain Size

Grain size measurements were also carried out, in order to firstly evaluate the

changes imposed by the friction process on the stud material and secondly to

investigate the influence of the welding parameters on the resultant microstructure,

by scanning the grain size distribution across the joint. Consequently, a relation

between microstructure and mechanical properties could be ascertained. The system

used to assess grain size was a LEICA DMI 5000 microscope with an image analysis

system (a4i-Analysis) interfaced to it. According to ASTM E-112-96 [134] grain size

was measured in 9 different positions across the transversal section, as illustrated

and indicated in Figure 3.7.

Figure 3.7: Position of grain size assessment across the welded joint.

3-51

3.5.2.3 Scanning Electron Microscopy/EDS

SEM/EDS was used in this investigation to characterise base materials and identify

second phase particles as well as when high resolution and high depth of field

images were necessary during the assessment of welded joints. The microstructural

analysis was performed using two types of imaging signals – back-scattered

electrons (BSE) and secondary electrons (SE).

Specimen preparation was identical to that used for macrostructural characterisation.

A Zeiss DSM 962 SEM was used during the analysis and the welded specimens

were not embedded, in order to have the appropriate electric contact between the

sample (cathode) and the SEM table (anode).

Energy dispersive spectroscopy (EDS) was performed to assess local chemical

composition along the joint line. Second phase particles were also investigated, in

the lamellae particles originally from the BM as well as in fractured particles in the

recrystallised zones of welded samples.

3.5.3 Hardness Testing

Microhardness testing using conventional Vickers hardness measurements was

performed on base material and across the joints at specific locations according to

DIN EN ISO 6507 [135] and DIN EN 1043-2 standards [136]. The measurements were

taken in three rows at 2 mm, 10 mm and 18 mm from the top surface and three

columns in the middle, at -4 mm and 4 mm from the stud central axle, as illustrated in

Figure 3.8. A load of 200p was applied for 30 s (HV0.2) with a distance between

indentations of 0.5 mm. The surfaces to be tested were polished with 4000 grit SiC

paper, which is a polishing paper with 3 μm average particle diameter, and etched, in

order to enable the correct identification of the different welding zones. Hardness

tests were performed on an EMCOTEST M1C010 testing machine with computer

controlled specimen positioning and indentation.

3-52

Figure 3.8: Position of hardness profiles on welded specimens.

3.5.4 Pull-out Testing

Pull-out tests were also performed in the welded joints based on the pull-out

standards found on the literature [137,138]. Pull-out specimens were produced from

welded samples with the rest of the stud machined similarly to the half of a “Form B”

(see DIN 50125 [138]) round tensile sample. Figure 3.9 shows the pull-out specimen

geometry and the position of the sample in relation to the joint. Figure 3.10 shows the

clamping system developed for the test as well as the complete assembly prior to

testing. Tests were performed with a deformation rate of 0.5 mm/min using a

Schenck-Trebel RM100 with 100 kN load capacity.

(a) (b)

Figure 3.9: Pull-out specimen geometry (a) and position of the sample in relation to the joint (b).

3-53

(a) (b)

Figure 3.10: Clamping system developed for the pull-out tests. In (a) the sample attached to the lower

fixture and in (b) the joint completely mounted on the machine prior to testing.

Considering the welded area as a function of the hole’s geometry, pull-out tests were

performed to calculate the magnitude of stresses acting directly on the joint, as the

samples were expected to collapse within the gauge length. Welded surface areas,

as is highlighted in red in Figure 3.9, were calculated in both cases using the 3D

mechanical design software SolidWorks and are presented below:

Ahole [6 mm, 20°] = 872.77 mm2

Ahole [8 mm, 20°] = 1029.35 mm2

3-54

3.5.5 Transversal Tensile Testing

Specimens for tensile testing were precision machined from the welded samples, as

well as from the base material, by spark erosion (EDM – electric discharge method)

with the welded interface stud/base plate located exactly in the centre of the tensile

specimen (see Figure 3.11). Specimens were first spark eroded and after lathed

according to DIN 50125 – B6 x 30 [138] (see Figure 3.11(a)).

Two specimens were machined from each joint and tested in laboratory air at room

temperature, according to the DIN EN 10002 standard [139]. The equipment used for

testing the samples was a screw-driven tensile testing machine Schenck-Trebel

RM100 with 100 kN load capacity. The traverse speed of 0.5 mm/min and the

displacement was recorded by a laser system coupled to the testing machine. Five

specimens were tested for each welding condition. Figure 3.11(b) represents the

position where the tensile specimen would theoretically intersect the stud and

illustrates both bonding lines between the base plate and the stud material.

(a) (b)

Figure 3.11: Tensile specimen geometry (a) and position of specimens relative to the joint (b).

3-55

3.5.6 Bolt-Load Retention Testing (BLR)

Bolt-Load retention tests were carried out, in order to simulate and further investigate

the loss of pre-load force used during the assembly of Mg transmission cases.

Transmission cases, typically joined by pre-loaded bolts, suffer from stress

relaxation, which could thus adversely impact the sealing of the joint. This issue has

been extensively reported on the literature [44-46,140-142]. A through-bolt type of bolt joint

configuration was used with a 30x30x30 mm hardened steel flange and a centred

10.5 mm diameter hole, according to DIN EN 20273 [143]. Figure 3.12 shows the

geometry of the BLR specimen (Figure 3.12(a)) and the position of the case relative

to the joint (Figure 3.12(b)). Furthermore, the threaded hole was positioned as

centred on the reinforced area.

The case was a 50x60 mm block, 18 mm thick and with the threaded hole concentric

with the block. The base materials and welded samples were tested for further

comparison at 125 °C and 175 °C, in accordance with ASTM E 328-86 [144].

Specimens were torqued with 40/60 Nm and 25/35 Nm, for steel and Al bolts

respectively, at room temperature and held one hour before being put in the oven.

This waiting period minimises the effects of short-term relaxation following the

tightening of bolted joints, as suggested by Bickford [145]. Tests were conducted over

periods of 100 hours, using Al and steel bolts fabricated according to DIN EN ISO

7989 [146] and DIN EN ISO 4014 [147] standards respectively. Ultrasonic measurement

equipment BoltMike III [148] was used to monitor the initial and remaining fraction of

load. With this equipment, an ultrasonic pulse, generated in the transducer, is

transmitted into the fastener. The pulse travels down the length of the fastener and is

reflected back from the end. The BoltMike III measures the total pulse transit time,

and based on the physical properties of the fastener material, calculates the tension

and clamp load according to ASTM E1685-00 [149].

3-56

(a) (b)

Figure 3.12: Schematic of BLR sample for tension. In (a) the geometry of the case and in (b) the

position of the case in relation to the welded sample.

In order to reproduce experiments carried

out by Westphal [35], some insights were

provided by the reinforced specimen being

positioned in the upper part of the joint. This

configuration would simulate the fastening of

an Mg gearbox directly into an Al (AlSi9Cu3)

engine block, as reported recently [6]. Figure

3.13 illustrates the configuration used, which

was tested at different temperatures and

with different bolt materials.

Figure 3.13: Schematic of BLR test with the

reinforced component in the upper part.

3.5.7 Creep Testing

Bearing in mind the main technological application addressed in this work and

knowing that joint performance is directly related to creep resistance, the creep

behaviour of welded samples in all three different material combinations was

evaluated. Welds were produced using the same procedure previously discussed.

Creep test samples with 6x15 mm were extracted from the extruded material

(denominated lower samples – see Figure 3.14) and from the upper stud area as a

reference (denominated upper or reference samples). Three temperatures (125 °C,

150 °C and 175 °C) and a constant compression load of 80 MPa were used.

3-57

Figure 3.14: Positions where creep samples were machined.

Creep results are presented in terms of creep curves and activation energy. In

general, the dependence of the minimum or steady state creep rate d/dtmin on stress

and temperature can be expressed for most metals as:

d/dtmin = A0.exp(-Qc/RT)σn, [Equation 3-1]

where A0 is a material constant, σ the applied stress, n the stress exponent, Qc the

activation energy for creep, R the gas constant and T the absolute temperature.

Based on Equation 3-1, in order to determine the activation energy, which represents

the necessary energy to activate the creep process, the minimum creep rate d/dt

should first be calculated [27]. To calculate d/dtmin one can define in the creep curves

the region with the minimum creep rate, from which the slope of the curve in the

selected interval gives the minimum creep rate directly. After d/dtmin has been

calculated, the activation energy can be determined by plotting ln(d/dtmin) versus

[T-1]. The linear curve fitting these data has the slope -Qc/R, from which the activation

energy can subsequently be calculated. For all the combinations, the activation

energy was calculated.

4-58

4 RESULTS

4.1 METALLURGICAL AND MECHANICAL CHARACTERISATION OF BASE

MATERIALS

4.1.1 Base Material AZ91D-T6

Table 4.1 shows the chemical composition of heat treated AZ91D.

Table 4.1: Chemical composition of the AZ91D ingot (T6 condition).

Chemical composition

Element Al Zn Si Cu Mn Fe Others Mg

Actual Composition

[wt.%] 8.5 0.63 0.04 0.005 0.22 0.013 --- Bal.

Nominal

Composition [14]

[wt.%]

8.5 9.5

0.45 0.90

0.05 0.015 0.17min 0.004 0.01 O.E. Bal.

Micrographs of the heat treated base material, shown in Figure 4.1, expose the

irregular structure dominated by the typical lamellae eutectic morphology surrounded

by isolated islands of a divorced eutectic structure, as described by [14]. The

microstructure is formed basically by a lamellar interdendritic Mg17Al12 precipitate (β-

phase) produced throughout the grains of Mg solid solution (α-phase) by artificial

aging. Particles of both Al-Mn (grey) and Mg2Si (blue) are also reported to be

typically found in Mg-Al-Zn alloys and this has been partially confirmed. Additionally,

some isolated islands of massive Mg17Al12 compound (white) were also identified.

EDS was performed to assess local chemical composition. Points are indicated in

Figure 4.1(b), (c) and (d) and identified as points 1 through 11. The corresponding

results are shown in Table 4.2. The average grain size varied from 200 μm up to

600 μm in both the central and the outer regions of the ingot.

4-59

(a) (b)

(c) (d)

Figure 4.1: Micrographs of AZ91D-T6 after heat treatment in (a) light microscope and EDS analysis of

the particles found in the base material (b), (c) and (d).

Table 4.2: Results of EDS analysis at 11 different points within the base material AZ91D-T6.

Mg(wt.%) Al(wt.%) Mn(wt.%) Zn(wt.%) Si(wt.%)

Point 1 4.80 50.64 44.56 --- ---

Point 2 62.89 31.48 --- 5.63 ---

Point 3 65.37 30.37 --- 4.27 ---

Point 4 61.68 32.42 --- 5.90 ---

Point 5 89.05 10.95 --- --- ---

Point 6 61.45 32.44 --- 6.11 ---

Point 7 1.35 36.47 58.87 --- 0.54

Point 8 2.10 40.1 55.37 --- 0.50

Point 9 2.51 43.51 48.84 --- 0.49

Point 10 2.30 40.44 56.59 --- 0.68

Point 11 2.28 36.74 58.94 --- 0.49

4-60

Table 4.3 lists some of the mechanical properties assessed during mechanical

characterisation, while Figure 4.2 presents tensile curves obtained for the BM.

Strength and elongation values ranging respectively from 140 MPa to 210 MPa and

from 0.5 % to 3 % were obtained, with the upper samples clearly presenting higher

strengths.

Table 4.3: Mechanical properties of the AZ91D ingot (T6 condition).

Mechanical Properties

Rm [MPa] Rp0,2 [MPa] A[%] Hardness

AZ91D-T6 (Tested) 140 - 210 100 - 140 0.5 – 3 65 – 80 HV

(a)

(b)

Figure 4.2: Stress-elongation curves for Mg AZ91D-T6 tensile samples. In detail, the longitudinal (a)

and transversal (b) direction.

4-61

4.1.2 Base Material AE42

Table 4.4 lists the chemical composition of the base material AE42, while Figure 4.3

presents the typical base material (BM) morphology, with selected regions for EDS

analysis.

Table 4.4: Chemical composition of the AE-42 ingot (cast condition).


Element Al Mn Si Ce La Ne Pr Others Mg

Actual Comp. [wt.%] 4.0 4.3

0.44 0.54

0.01 1.3 0.61 0.64

0.38 0.42

0.14 Zn

0.003 Bal.

Nominal Comp. [8]

[wt.%] 3.6 4.4 0.27 0.01 2.0 – 3.0 RE --- Bal.

(a)

(b) (c)

Figure 4.3: Micrographs of AE42 (a) and points for EDS analysis (b) and (c).

4-62

The ingot’s material is generally characterised by a pronounced dendritic structure in

the matrix, in combination with dispersed lamellae and particulate second phase

particles of Al11RE3, preferentially located along the grain boundaries [29,150,151]. Figure

4.3(b) and (c) show backscattered electron images of the BM, evidencing second

phase particles with different sizes, randomly distributed within the Mg matrix. Points

subjected to EDS analysis are indicated and identified as points 1 through 6. The

corresponding results are shown in Table 4.5.

Table 4.5: Results of EDS analysis at six different points within the base material AE42HP.

Mg

(wt.%)

Al

(wt.%)

Si

(wt.%)

Mn

(wt.%)

RE

(wt.%)

Point 1 98.80 1.20 --- --- ---

Point 2 98.37 1.63 --- --- ---

Point 3 0.32 34.47 --- 41.25 23.96

Point 4 0.55 32.24 1.38 38.59 25.48

Point 5 4.45 33.32 --- 0.60 59.63

Point 6 4.04 34.33 --- 0.68 58.47

Table 4.6 presents some relevant mechanical properties of the base material AE42,

while Figure 4.4 presents the tensile curves recorded during material

characterisation. In the longitudinal direction, the samples investigated presented

tensile strengths varying between 120 MPa and 130 MPa, with elongations values

ranging from 5 % to 7 % in the upper part of the ingot. On the other hand, in the

lower areas, similar strengths of around 90 MPa were observed, however with the

elongation varying between 3 % and 6 %. In the transversal direction, tensile strength

and elongation presented values varying between 85 MPa and 130 MPa and

between 3 % and 7 % respectively, independent of the position in relation to the

ingot.

Table 4.6: Mechanical properties of the AE-42 ingot (cast condition).



AE-42 (Tested) 85 - 130 --- 3 - 7 39 - 49

4-63

(a)

(b)

Figure 4.4: Base Material tensile curves for longitudinal (a) and transversal (b) round AE42 samples.

Ingot detail indicates the position where the samples were machined from.

4.1.3 Base Material MRI230D

Table 4.7 lists the actual MRI230D chemical composition.

4-64

Table 4.7: Chemical composition of the MRI230D ingot (cast condition).


Element Al Zn Si Cu Mn Fe Others Mg

Actual Composition

[wt.%]

6.6 7.5

0.002 < 0.01 0.002 0.29 0.33

--- Ca: 2.1 – 2.4

Sr: 0.29 – 0.33 Sn: 0.71 – 0.89

Bal.

Nominal

Composition [33]

[wt.%]

6.4 7.2 0.08 0.03 0.005

0.17 0.40 0.004

Ca: 1.8 – 2.6 Sr: 0.4

Bal.

Micrographs of the MRI230D base material (see Figure 4.5) indicate microstructures

consisting of the α-Mg matrix and intermetallic compounds that form an almost

continuous network distributed at grain boundaries and interdendritic areas. The

morphologies of the intermetallic compounds in these alloys have been revealed by

backscattering electron SEM images.

Figure 4.5: Micrographs of MRI230D (above) and backscattering electron SEM images (below) with

points identified for further EDS analysis.

4-65

As shown in Figure 4.5, the interdendritic compounds show three morphologies: thin

lamellar, a coarse irregular shaped eutectic and a bulky phase located at triple

junction of grains. Stable, round Al-Mn based precipitates were also observed to

compose the microstructure of such alloys. In order to identify the local chemical

composition of these intermetallic compounds existing in the as-cast microstructure,

EDS was performed. Points are indicated in Figure 4.5 and identified as 1 to 13. The

corresponding results are shown in Table 4.8. Additional microscopic observations

revealed an average grain size varying between 90 μm to 120 μm in the centre and

in the periphery of the ingot.

Table 4.8: Results of EDS analysis at 13 different points within the base material MRI230D (in wt.%).

Mg Al Ca Mn Zn Sr Sn

Point 1 --- 27.15 --- 42.18 15.75 --- 7.67

Point 2 7.23 --- 6.25 --- --- --- 86.52

Point 3 16.61 --- 11.51 --- --- 2.14 69.73

Point 4 12.31 51.93 34.99 --- --- --- ---

Point 5 --- 19.74 --- 43.40 24.09 --- 10.13

Point 6 46.36 35.21 18.43 --- --- --- ---

Point 7 96.35 3.65 --- --- --- --- ---

Point 8 96.86 3.15 --- --- --- --- ---

Point 9 96.57 3.43 --- --- --- --- ---

Point 10 --- 26.79 --- 47.07 13.96 --- 4.58

Point 11 --- 27.59 --- 43.47 14.45 --- 5.59

Point 12 --- 9.63 --- 19.46 38.39 --- 11.43

Point 13 --- 22.70 --- 39.18 21.29 --- 5.68

Table 4.9 presents further mechanical properties, while Figure 4.6 shows stress-

strain curves, plotted during tensile testing. In both cases, i.e. in the longitudinal and

transversal directions, the upper samples (1 to 5) presented strength and elongation

values varying between 100 MPa and 120 MPa and 1.5 % and 2 %, respectively. In

the lower part of the ingot, elongation values were similar and ranged between 1 %

and 1.5 % in both directions, while tensile strengths were around 90 MPa and varied

between 60 MPa and 90 MPa for transversal and longitudinal directions respectively.

4-66

Table 4.9: Mechanical properties of the MRI230D ingot (cast condition).



MRI230D (Tested) 60 - 120 50 - 100 0.5 – 2 46 – 54 HV

(a)

(b)

Figure 4.6: Base Material tensile curves for longitudinal (a) and transversal (b) round MRI230D

samples. Positioning of samples in relation to the ingot is shown in the detail (right).

4-67

4.2 SIMILAR AZ91D-T6 JOINTS – 6 MM DIAMETER/12 MM UPSETTING

As the rotational speed has been demonstrated to be the least sensitive parameter in

friction welding [73,118,152] and considering the experience acquired from previous

trials, FHPP joints were performed in this section in a similar configuration, varying

the welding pressure and forging force. On the other hand the upsetting, as one of

the main welding parameters, has shown to play an important role in rotational

friction welding. Therefore, its influence on FHPP is also investigated in this study

with the respective results presented in Section 4.3. In FHPP, a certain amount of

upsetting is required to fill the hole completely. If the upsetting is too small, the hole

will not be filled up to the surface, while if it is too high, much stud material is

plasticised on top of the weld, heating the consumable stud and the base plate

excessively, and needlessly extending the welding time in most of cases. For the

experiments performed within this section, all welding studs and base plates were

machined from the base material AZ91D-T6. The chemical composition and

mechanical properties of this base material have been reported previously (see

Section 4.1.1). Based on previous studies carried out by the author and on some

prior investigations on FHPP [78,82,99,100], a combination involving 10° tapered studs

with 6 mm tip diameter and base plates with 20° tapered holes and 6 mm bottom

diameter was chosen. Within this configuration, even minor bonding defects,

especially at the wall of the holes, would not be expected, due to the horizontal

component of the forces created by the taper angle. A parameter matrix was devised

and extended to the operational limits of the system on one side and to the

mechanical properties of the materials on the other. The rotational speed was kept

constant at 4000 rpm while the axial pressure was changed from 6.90 to 13.79 bar in

3.45 bar steps. The upsetting was set to 12 mm, based on early trials, and the

forging pressure was always 13.79 bar higher than the welding pressure. Nine welds

for each parameter combination were produced and designed for a certain test, as

listed in Table 4.10. In this context, welds “F” to “I”, for example, were used only for

transversal tensile testing. Throughout this section the welds will be referred to as 1,2

or 3, according to the set of parameters used.

4-68

Table 4.10: Designation of welded samples (AZAZ-06/12 series) according to the testing program.

Weld Welding

Parameters (bar x rpm)

Process Stability

Temperature

Met. Analysis

Hardness

Pull-out Test

Transversal Tensile Test

AZAZ 1A 1I

6.90 x 4000 12 mm Ups.

FP: 20.69 bar 1A 1E 1A 1E 1C 1C

1A, 1B 1C, 1E 1F 1I

AZAZ 2A 2I

10.34 x 4000 12 mm Ups.

FP: 24.13 bar 2A 2E 2A 2E 2C 2C

2A, 2B 2C, 2E 2F 2I

AZAZ 3A 3I

13.79 x 4000 12 mm Ups. FP: 27.58

3A 3E 3A 3E 3C 3C 3A, 3B 3C, 3E 3F 3I

Figure 4.7 presents the process diagram, including the most important parameters

and additional information about the process, such as welding duration and holding

time (6 s for every weld). As can be seen, the rotational speed is started first and

after the set speed is reached (in this case 4000 rpm), the stud is moved down,

making contact with the bottom of the cavity. From this point the axial downward

movement is measured as upsetting. As soon as both faying surfaces are brought in

contact, the rotational speed falls briefly, due to the first initial friction being corrected

by the control system almost instantaneously. Simultaneously the welding pressure

reaches its preset value and the welding stud starts to be deposited. When the

predetermined upsetting distance is reached (the welding time is finished), the

rotational speed is stopped and the pressure increased to the preset value. Finally,

the forging force is held constant for 6 s (forging time) to consolidate the joint.

4-69

Figure 4.7: Welding progress diagram (welding pressure 13.79 bar, forging pressure 27.6 bar,

rotational speed 4000 rpm and 12 mm upsetting).

4.2.1 Process Monitoring

As mentioned on the experimental procedure, data files were stored for every weld

produced in this study. These files were then transferred to a standard data

processing software (EXCEL) for further processing and complementary evaluation.

Figure 4.8 presents the upsetting curves generated by the data acquisition system.

Based on this and on the calculated upsetting rate (see in Figure 4.8, UR values for

the different welding conditions), the influence of the welding pressure on the welding

time can be clearly observed. Welds performed with low pressures present on

average a welding time (WT) of 24.4 seconds (s), whereas high pressure joints are

consolidated in around 9.5 s. Furthermore, the total upsetting (TU), which mirrors the

total shortening of the welded pair, increases from 15.5 mm to 16.3 mm with

increasing forging pressure.

4-70

Spec. WT ST TWT UWT UST TU

1A 27.2 0.8 28.0 13.0 2.4 15.4

1B 23.9 0.9 24.8 13.6 2.3 15.9

1C 29.5 0.9 30.4 13.7 2.4 16.1

1D 19.1 1.0 20.1 12.8 2.4 15.2

1E 22.0 1.0 23.0 12.7 2.4 15.1

AV 24.4 0.9 25.3 13.2 2.4 15.5

SD 4.1 0.1 4.0 0.5 0 0.4

(a) 6.90 bar x 4000 rpm - UR: 0.541 mm/s.


2A 14.9 0.9 15.8 13.0 3.0 16.0

2B 17.2 0.9 18.1 13.4 2.7 16.1

2C 17.0 1.0 18.0 12.8 3.3 16.1

2D 17.2 0.9 18.1 12.7 3.3 16.0

2E 12.8 1.0 13.8 13.1 3.5 16.6

AV 15.8 0.9 16.7 13.0 3.2 16.2

SD 2.0 0.0 1.9 0.3 0.3 0.2

(b) 10.34 bar x 4000 rpm - UR: 0.823 mm/s.


3A 9.4 1.0 10.4 12.7 3.0 15.7

3B 8.8 0.9 9.7 12.7 3.4 16.1

3C 9.9 1.3 11.2 12.9 3.6 16.5

3D 9.0 1.0 10.0 13.0 3.2 16.2

3E 10.7 1.1 11.8 12.9 3.5 16.4

AV 9.5 1.0 10.6 12.8 3.3 16.3

SD 0.8 0.1 0.9 0.1 0.5 0.6

(c) 13.79 bar x 4000 rpm - UR: 1.347 mm/s.

Figure 4.8: Process stability and the influence of axial pressure on the upsetting rate (WT: welding

time; ST: stopping time; TWT: total welding time; UWT: upsetting during the welding time;

UST: upsetting during the stopping time; TU: total upsetting and UR: upsetting rate).

4-71

4.2.2 Thermal Cycle

Figure 4.9 shows the measurements of three sets of joints, consisting of five samples

per set, welded with different axial pressures and monitored in two different positions.

Although the axial pressure significantly influences the welding time (see Section

4.2.1), this tendency cannot be directly transferred to the maximum temperature.

While welding time showed a significant variation by duplicating the pressure, the

peak temperatures presented a variation of 16 °C (≈4 %) in T1 and 33 °C (9 %) in T2

under similar conditions, i.e. between samples AZAZ-1 and AZAZ-3.

For low pressure welds (AZAZ-1), average peak temperatures of 427 °C and 361 °C

respectively were measured in upper (T1) and lower (T2) weld areas. For high

pressure welds, however, average peak temperatures of 411 °C and 328 °C

respectively were observed in upper and lower weld areas. Extending the analysis

and considering the cooling rates demonstrated in Figure 4.9, one can observe a

longer cooling time for welds produced within the AZAZ-1 series. After 40 seconds,

the upper thermocouples indicated temperatures around 195 °C and 95 °C

respectively for AZAZ-1 and AZAZ-3 series.

4-72

Cooling Rate (°C) Cooling Rate (°C) (a)

Peak Temperature 40s 50s 60s

(b) Peak

Temperature 40s 50s 60s Average 427 °C 194 148 121 Average 361 °C 182 142 119

AZAZ-1C 448 °C 249 181 147 AZAZ-1C 383 °C 231 175 144

Cooling Rate (°C) Cooling Rate (°C) (c)


(d) Peak


AZAZ-2C 389 °C 142 115 97 AZAZ-2C 368 °C 139 113 97

Cooling Rate (°C) Cooling Rate (°C) (d)


(e) Peak


AZAZ-3C 395 °C 91 77 67 AZAZ-3C 281 °C 87 74 66

Figure 4.9: Thermometric results for similar AZAZ welds. Measurements respectively at 6 mm (T1) and

17 mm (T2) from top surface for 6.90 bar x 4000 rpm (a) and (b), 10.34 bar x 4000 rpm (c) and (d) and

13.79 bar x 4000 rpm (d) and (e) welds.

4-73


Weld-zone cross-sections indicate joints with no porosity and without any lack of

bonding along the welding line in all the investigated conditions. At the bottom of the

stud, under higher magnification, no clearly visible transition zone (bonding line) was

identified, suggesting an intimate contact between the stud/base plate pair, imposed

mostly by the forging pressure. A clear bonding line could only be observed in the

sidewall interface, since the forces acting in those areas are not directly applied by

the process, but are transferred by a combination of solid and hydrostatic conditions

via the plasticised material. The influence of the welding parameters on the material

flow is shown in Figure 4.10. Based on a visual analysis, it can be clearly seen that

joint 1C, welded with lower pressure, and therefore with a longer welding time,

presented a larger heat affected zone (HAZ) when compared with weld 3C.

(a) AZAZ-1C (6.90x4000).

4-74

(b) AZAZ-2C (10.34x4000).

(c) AZAZ-3C (13.74x4000).

Figure 4.10: Friction welding macrostructures of similar AZ91D-T6 x AZ91D-T6 joints.

4-75

Figure 4.11 evidences, in terms of microstructural changes, the difference in the

global heat generation. While in sample 1C the microstructure in regions adjacent to

the welding area is completely recrystallised, with most of the eutectic phases

disrupted and entirely dissolved into the Mg-Al solid solution (see Figure 4.11(a), in

sample 3C the typical lamellar eutectic morphology, formed during the ingot casting

process, is observed in an extension very close to the BL (see Figure 4.11(b).

(a) Microstructure in region A – see Figure 4.10a. (b) Microstructure in region B – see Figure 4.10c.

Figure 4.11: Comparison of microstructures at the same position between samples 1C and 3C.

As shown in Figure 4.12, the microstructure of FHPP welded specimens generally

consists of a homogenised area in the lower part of the consumable and a region of

plastically deformed stud material in the upper part. Figure 4.12(a) and (b) comprise

the microstructures around the side bonding line, which is identified by means of a

dashed line. The grain size can be seen to decrease in the vicinity of the joint in an

extension varying between 50 μm and 150 μm. Even under higher magnification it

can be observed that the lateral interface presents a smooth transition between stud

and base plate material (Figure 4.12(b)). The recrystallised Mg grains in a region

immediately adjacent to the bonding line (see Figure 4.12(c)) can be seen to be fine,

homogeneous and equiaxed. The frictional heat created by the rubbing of both faying

surfaces and the mechanical working imposed by the process provided the driving

force for dynamical recrystallisation of the materials within the weld zone. Figure

4.12(d) shows the transition area at the bottom of the cavity on the right side,

indicating no visible discontinuity on the interface.

4-76

The lamellar eutectic phases Mg17Al12 seem to be dissolved and re-precipitated as

large particles with various shapes around the primary-grown Mg-Al solid solution.

However, the coarse second phase particles, even those exactly over the bonding

line, are not mechanically fragmented and remain in the composition of the

microstructure (see indications in Figure 4.12(a) and (d). Three main weld zones, in

agreement with the literature [129,153-155], can be pointed out according to the type and

intensity of the supplied energy (heat input) and deformation they were subjected to

during the welding cycle:

HAZ – heat affected zone, where the microstructure and respective

mechanical properties only suffer the influence of the heat generated during

the welding process. At lower magnifications, such regions did not show any

significant microstructural change in comparison with the base material (BM),

but grain growth can still be observed, depending on the energy input used to

perform the weld;

TMAZ – thermo-mechanical affected zone, where the amount of heat is

higher than in the HAZ and grains experience a substantial deformation,

essentially resulting in recovery of the highly strained grains. In this region, the

pancake grains from the BM are deformed and can be observed easily in the

micrographs surrounding the joint line (see Figure 4.12(c)), as a transitional

interface between HAZ and the recrystallised zone, where grains are equiaxed

and refined. Furthermore, a rearrangement of dislocation and formation of

sub-grain interfaces is expected in this region, since recovery takes place;

RZ – recrystallised zone is distinctly characterised by the formation of a fine

grain microstructure with equiaxed morphology as a result of the dynamic

recrystallisation phenomenon. The refined microstructure of the recrystallised

zone is the core of the joining among the two workpieces and corresponds to

the microstructure resulting from the highest levels of deformation and

temperature during the friction welding process. In FHPP, as the consumable

rod undergoes significant hot working and is therefore fully plasticised across

the entire diameter, the total filled up area can considered to be the

recrystallised zone.

4-77

(a) Sidewall recrystallised zone. (b) Lateral transition area without

apparent discontinuity.

(c) Microstructural development at the lower

stud areas with indication of welding zones.

(d) Transition area.

Figure 4.12: Macro and micrographs of the AZAZ-1C joint.

HAZ

TMAZ

RZ RZ

RZ RZ

RZ BM

Stud

BM

Stud

4-78

The average grain size within the deposited (stud) material varies between 7 μm in

the lower regions of the joint to 17 μm in the upper parts. Three measurements were

performed in each position, for each sample, i.e. the results presented in Figure 4.13

are the average of three scans. Base material has, on the other hand, a grain size

varying between 200 μm and 600 μm.

Figure 4.13: Grain size distribution within the deposited material (values in μm).

As mentioned above and demonstrated in Figure 4.14(a) and (b), AZ91D-T6 cast

ingots have a grain size varying between 200 μm and 600 μm. Their typical

microstructure is basically formed by an α-Mg matrix, a lamellar interdendritic α-Mg-

Mg17Al12 precipitate (β-phase) and some isolated islands of Mg17Al12 compound.

Similar to the bonding line, the microstructure within the deposited material is

basically characterised by the complete dissolution of the typical base material

phases, followed by the formation of a very fine, equiaxed and dynamically

recrystallised structure. The previous Mg(Al,Zn) phases are either dissolved into the

α-Mg matrix or precipitate at the grain boundaries, according to Figure 4.14(c) and

(d). Several EDS analysis were performed in this region, in order to identify the

resultant phase compositions after welding. Figure 4.14 and Table 4.11 indicate and

show respectively the 10 different points where such EDS analyses were performed.

4-79

(a) (b)

(c) (d)

Figure 4.14: Microstructural characteristics of the AZ91D-T6 BM (a) and (b) and of the extruded

material within the welding zone (c) and (d).

Table 4.11: Results of EDS analysis in 10 different points within the extruded material.

Mg(wt.%) Al(wt.%) Mn(wt.%) Zn(wt.%)

Point 1 87.67 10.97 --- 1.36

Point 2 85.19 12.98 --- 1.34

Point 3 87.16 11.62 0.12 0.42

Point 4 87.18 11.77 0.12 0.93

Point 5 87.19 11.25 0.13 1.04

Point 6 90.68 8.59 --- 0.88

Point 7 87.39 11.18 0.14 1.28

Point 8 92.35 6.56 --- ---

Point 9 93.70 6.03 --- ---

Point 10 93.56 6.14 --- ---

Lamellar Mg17Al12 precipitate

α-Mg

Mg17Al12 compound

4-80

4.2.4 Mechanical Characterisation

4.2.4.1 Hardness Tests

In the case of similar AZ91D-T6 joints, microhardness profiles were performed along

the transversal cross section surfaces for the three different conditions, using the

procedures listed in Section 3.5.3. Figure 4.15 presents the microhardness profiles in

welds produced with low (1C), intermediate (2C) and high (3C) pressures at 2 mm,

10 mm and 18 mm from the top plate surface. Hardness values in the BM were

shown to vary from 60 HV to 120 HV in cases where the indenter hit a harder phase

or precipitate. In contrast to the BM scans, hardness profiles within the extruded

material point out similar values of approximately 80 HV for all three welded

conditions at 2 mm and 10 mm from the top surface. At 18 mm, however, a slight

increase in the average hardness values was observed. Microhardness profiles

indicated in those areas values over 80 HV for the three welding conditions.

Further analysing Figure 4.15(a) to (c), a significant decrease in the scattering within

the extruded material was observed in comparison with the BM. Furthermore, profiles

performed 2 mm away from the top surface presented a wider region outside the

extruded zone, where a reduced dispersion in hardness was observed. On the other

hand, measurement of scattering in the lower stud areas (see Figure 4.15(c))

increases significantly, as soon as the scan crosses the bonding line and reaches the

BM, inside of the TMAZ/HAZ.

(a)

4-81

(b)

(c)

Figure 4.15: Hardness survey of welded samples, transversal to the consumable at 2 mm (a),

10 mm (b) and 18 mm (c) from the top surface, according to Figure 3.8.

4.2.4.2 Pull-out Tests

Pull-out properties were determined at room temperature for the AZ91D-T6 alloy for

the three different welding conditions. Based on Figure 4.16, pull-out tests confirm

the suitability of the process to perform high strength welds. Generally, similar joints

achieved BM strength values of 35 kN for low (AZAZ-1 series), intermediate (AZAZ-2

series) and high (AZAZ-3 series) pressures with no failure observed within the

welded area (see corresponding detail in Figure 4.16).

4-82

Figure 4.16: Pull-out test results for similar MgAZAZ-06/12 welds.

4.2.4.3 Transversal Tensile Tests

According to Table 4.10, four samples were welded in the same condition to evaluate

the bonding strength at the sidewall. Two tensile samples were machined from each

specimen (see Figure 3.11). Figure 4.17 presents the tensile strength results,

considering the average strength obtained from four specimens tested at the same

location. According to this, joints performed under different welding conditions

generally show similar mechanical properties to those of the base material. Low

pressure joints (AZAZ-1 series) achieved strength values ranging from 150 MPa up

to 167 MPa, for lower and upper weld regions respectively. High pressure welds

(AZAZ-3 series), however, presented strength values comparable to those from the

BM, achieving 100 % joint efficiency in lower and upper weld areas. A significant

decrease in the elongation was observed for low and high pressure welds, reaching

values inferior to 1 %. BM elongation, on the other hand, presented values ranging

from 0.5 % to 3 % according to Table 4.3. Low, intermediate and high pressure welds

also showed equivalent yield strengths, indicating values of 90 % to 100 %, 86 % to

100 % and 97 % to 100 % of the BM properties for lower and upper weld areas

respectively.

4-83

Figure 4.17: Transversal tensile test results. Values in brackets indicate the weld efficiency with

reference to the base material (BM) for low AZAZ-1, intermediate AZAZ-2 and high AZAZ-3 welding

pressures (see Figure 3.11 and Table 4.10 for details).

4.3 DISSIMILAR JOINTS – 8 MM DIAMETER/10 MM AND 20 MM UPSETTING

In Section 4.2, similar AZ91D-T6 welds with 6 mm diameter and 12 mm upsetting

(AZAZ-06/12 series) were presented. By this approach, the influence of the welding

pressure on the process as well as mechanical properties and the joint formation

mechanism for similar welds (see Section 5.2.2) were established. As a further stage,

dissimilar welds, using higher temperature resistant consumable materials were

performed and will be assessed in this section. Further, similar joints, using the same

geometry utilised for the dissimilar welds, were also performed for comparison

purposes. Hence, all the results of the so-termed “8 mm diameter/10 mm and 20 mm

upsetting” welds will be presented in this section.

4-84

For the following experiments studs were machined from base material AE42,

MRI230D and AZ91D-T6, while base plates were produced from base material

AZ91D-T6. Chemical composition and mechanical properties of these base materials

have been previously reported (see Section 4.1). A geometry combination was

chosen of an 8 mm tip diameter 10° tapered stud and an 8 mm bottom diameter 20°

tapered hole. With this configuration, larger bolts could be used to perform the small-

scale component tests. After some initial trials, one parameter set (i.e. rotational

speed and pressure), was selected as the basis to perform the welds. Rotational

speed and welding pressure were kept constant at 4000 rpm and 13.79 bar

respectively, while 10 mm and 20 mm upsetting were used. Forging pressure was

always set to 17.24 bar. As in the previous case, nine welds for each parameter

combination were produced and devised for a specific test, as listed in Table 4.12.

Table 4.12: Designation and allocation of welded samples.

Weld Welding

Parameters Process Stability

Temperature

Met. Analysis

Hardness

Pull-out Test

Transversal Tensile Test

AE, MRI, AZ

with AZ

(1, 2, 3…, 9)

4000 rpm

WP: 13.79 bar

FP: 17.24 bar

10 mm Ups.

1 – 5 1 – 5 4 4 1, 2, 3, 5 6, 7, 8, 9

AE, MRI, AZ

with AZ

(a, b, c…, h)

4000 rpm

WP: 13.79 bar

FP: 17.24 bar

20 mm Ups.

a – d a – d a a b, c, d e, f, g

4.3.1 Process / Temperature Monitoring

Similar to the analyses previously performed, the welding process was monitored for

five joints within this investigation. Based on data delivered by the DAS it could be

observed that the weld progress follows its particular deposition path for each

material combination, however, with some instability within the same group. Welding

process curves, in terms of upsetting/upsetting rate, were obtained and kept for

further evaluation. Figure 4.18 presents the total welding time versus the achieved

temperatures in upper (Figure 4.18(a)) and lower (Figure 4.18(b)) weld areas.

Average peak temperatures with a respective scattering of results are summarised in

Figure 4.19.

4-85

The peak temperatures measured for the AEAZ joints ranged between 453 °C to

470 °C and 492 °C to 524 °C in upper and between 406 °C to 442 °C and 462 °C to

476 °C in lower weld areas for 10 mm and 20 mm welds respectively. Welding times

varied from 9.65 s to 15.72 s and from 31.9 s to 46.53 s for 10 mm and 20 mm

upsetting welds respectively. For MRIAZ 10 mm upsetting joints, temperature

indications ranging from 453.8 °C to 484.4 °C and from 398.8 °C to 421.8 °C in upper

(T1) and lower (T2) weld areas were observed, with corresponding welding times

varying between 19.08 s and 23.31 s. For higher upsetting, values from 485 °C to

495 °C and 389 °C to 446 °C were measured for T1 and T2 respectively. Welding

times varied in these cases between 19.1 s and 23.3 s, and between 44.6 s and

65.2 s respectively for 10 mm and 20 mm upsetting. Moreover, the AZAZ joints

experienced temperatures ranging between 430 °C to 464 °C and 467 °C to 497 °C

in the upper and between 361 °C to 381 °C and 390 °C to 414 °C in the lower weld

areas, for 10 mm and 20 mm welds respectively. Welding times varied from 20.4 s to

28.5 s and from 30.1 s to 51 s, for 10 mm and 20 mm upsetting welds respectively.

(a)

4-86

(b)

Figure 4.18: Peak temperatures in upper (a) and lower (b) thermocouples for similar and dissimilar

welds, 10 mm and 20 mm upsetting.

Figure 4.19: Average peak temperature in upper (T1) and lower (T2) weld areas for AEAZ, MRIAZ and

AZAZ, 10 mm and 20 mm upsetting welds.

4-87


4.3.2.1 AEAZ Joints

Macrostructural investigations of the dissimilar AEAZ joints indicate that sound welds

with no defects, such as porosity or lack of bonding, can be produced using FHPP

(see Figure 4.20). Since both materials are composed of different alloying elements

and therefore have different microstructural features, in most of the cases the

bonding line could be clearly identified suggesting an intimate contact between

consumable and base plate. As a consequence of the higher heat generation, which

can be estimated through the width of the heat affected zone, in the upper joint

regions (see Figure 4.20(a) and (e)) a smooth transition between stud and base

material can be seen. However, in the lower parts of the joint, where the heat is

almost instantaneously conducted out of the weld region, the border between

consumable and base plate is well defined. On the AZ91 side of the weld, a thin layer

of dynamically recrystallised grains presenting a non-uniform thickness extending up

to 150 μm from the welding line followed by an almost unmodified base material was

observed (see Figure 4.20(b) and (d)). In the central region at the bottom of the

cavity, where a higher amount of plasticised material is usually concentrated (see

Figure 4.20(c)), it seems that the consumable is mixed with the base material.

EDS was performed to assess the local chemical composition along the joint line.

Second phase particles were also investigated within the RZ. Points chosen for EDS

analysis are shown in Figure 4.21, and the corresponding results are presented in

Table 4.13.

4-88

(a)

(e)

(b)

(c

)

(d)

Fig

ure

4.20

: Mic

rost

ruct

ural

det

ails

of t

he d

issi

mila

r A

E42

/AZ

91D

-T6

join

t afte

r et

chin

g in

a s

olut

ion

with

20

ml d

istil

led

wat

er, 1

00 m

l eth

anol

, 6 m

l gla

cial

acet

ic a

cid

and

12 g

Pic

ric A

cid.

In th

e ce

ntre

on

the

left

side

an

over

view

of t

he jo

int c

ross

sec

tion;

In (

b), (

d) a

nd (

e) m

icro

grap

hs ta

ken

with

OM

; In

(a)

and

(c)

BS

E s

cans

det

ailin

g th

e bo

ndin

g lin

e.

4-89

Figure 4.21: Weld interface of the dissimilar AE42/AZ91D-T6 FHPP welded joint. Backscattered

electron micrograph showing points for EDS analysis in particles distributed along the joint line.

Table 4.13: Results of the EDS analysis in ten different points along the BL (wt.%).

Point Mg Al Zn Mn Ce Nd Pr La

1 8.94 34.89 --- --- 31.81 7.35 --- 17.01

2 4.51 34.33 --- 0.68 31.81 8.76 4.06 13.84

3 96.27 3.73 --- --- --- --- --- ---

4 98.16 1.84 --- --- --- --- --- ---

5 92.97 7.03 --- --- --- --- --- ---

6 92.86 7.14 --- --- --- --- --- ---

7 93.06 6.94 --- --- --- --- --- ---

8 92.82 7.18 --- --- --- --- --- ---

9 88.51 10.68 0.81 --- --- --- --- ---

10 88.40 10.59 1.01 --- --- --- --- ---

4-90

4.3.2.2 MRIAZ Joints

As for the previously investigated AEAZ joints, micrographs of dissimilar MRIAZ

welds suggest that sound joints can be produced using FHPP. Joints were

characterised by longer welding times (see Figure 4.18), which led the

microstructures to present wider HAZs (see overview in Figure 4.22). Figure 4.22(a)

details, in upper weld areas on the left-hand side, the transition between the AZ91

base material, where a narrow region with grain growth can be observed, and the

MRI consumable with a significantly refined microstructure. Macrographs from the

bottom regions of the joint, i.e. Figure 4.22(b), (c) and (d), show a severely deformed

area in the vicinity of the joint on the AZ91 side of the weld, where dynamically

recrystallised grains appear in an extension from 100 μm to 400 μm from the weld

line without, however, the presence of the eutectic structure. In those regions no

voids or discontinuities along the joint line, referred to as lack of bonding, were

observed. On the MRI side of the weld, a full degradation of the typical base

material’s second phase Mg(Al,Ca) particles was observed. Figure 4.22(e) presents a

micrograph obtained from the upper joint parts, where a smooth transition can be

observed, probably due to the role played by the forging force. In those areas an

exact determination of the bonding line was also not possible.

4-91

(a

)

(e

)

(b)

(c

)

(d

)

Fig

ure

4.22

: Mic

rost

ruct

ural

det

ails

alo

ng t

he b

ondi

ng li

ne o

f the

dis

sim

ilar

MR

I230

D/A

Z91

D-T

6 w

eld

(a),

(b)

, (c)

and

(e)

. In

(d)

a S

EM

mic

rogr

aph

show

ing

clus

ters

of p

artic

les

in r

egio

ns a

djac

ent t

o th

e B

L as

wel

l as

diffe

rent

wel

d zo

nes.

4-92

EDS analyses along the welding line indicate the presence of diverse eutectics with a

chemical composition varying in a fairly large range. Points for EDS analyses are

shown in Figure 4.23, and the corresponding results are presented in Table 4.14.

Figure 4.23: Points for EDS analysis - MRI230D/AZ91D-T6 weld (see Figure 4.22, position “a” for

reference).

Table 4.14: Results of the EDS analysis over the bonding line (wt.%).

Point Mg Al Zn Mn Ca Sr

1 10.32 45.85 --- 43.83 --- ---

2 33.76 35.67 0.99 26.39 1.96 0.40

3 34.33 41.97 --- --- 22.62 1.07

4 36.68 39.91 --- --- 22.74 0.67

5 65.06 25.46 --- --- 9.12 0.36

6 83.29 13.06 --- --- 3.01 0.45

7 93.99 6.01 --- --- --- ---

8 93.51 6.49 --- --- --- ---

9 73.19 19.62 --- --- 7.19 ---

10 83.31 14.11 --- --- 2.59 ---

11 57.98 17.92 --- --- 1.18 10.04

4-93

4.3.2.3 AZAZ Joints

Also within this geometry, since the process has shown to be suitable for performing

similar defect-free welds with 6 mm diameter consumables, Friction Hydro Pillar

Processing demonstrates its feasibility to produce visually high quality welds without

apparent discontinuities along the joint line with 8 mm diameter consumables. Figure

4.24 comprises not only an overview of the MgAZAZ-4 weld, but also some

micrographs taken along the bonding line. Figure 4.24(a) and (e) show, on the

consumable side, the typical refined, equiaxed and homogeneous microstructure

resultant from the dynamic recrystallisation (DRX) processes, which take place

during welding. The base plate, however, is seen to have grains of unequal size and

distribution intercalated with islands of non-recrystallised material in regions near to

the bonding line. On the other hand, micrographs from the bottom of the weld (see

Figure 4.24(b), (c) and (d)) show a broad range on both sides, where a dynamically

recrystallised structure is observed. Figure 4.24 shows also within the recrystallised

zone at the bottom, intercalated microstructures with lamellar-like shear bands.

Additionally, in the lower consumable stud areas, where the most severe hot working

is imposed, AlMn and MgSi (assumed to be Al8Mn5 and Mg2Si respectively) second

phase particles are not split and remain still composing the microstructure. Figure

4.25 shows the presence of AlMn (dark grey precipitates with black indications) and

MgSi (green compounds with white indications) phases over the bonding line.

Welds with higher upsetting generally show larger grain sizes within the extruded

area, larger recrystallised and heat affected zones with a much reduced grain size

gradient along the transitioning zone and an apparently much more homogeneous

structure in the deposited material. Grain size measurements in 10 mm upsetting

welds indicate average diameter values ranging from 9 μm in the lower to 20 μm in

the upper weld areas respectively, whereas the 20 mm welds show indications

varying between 8 μm in the lower and 28 μm in the upper areas. EDS was

performed to assess the local chemical composition. Points are indicated in Figure

4.26, taken in different areas along the welding line. The corresponding results are

shown in Table 4.15.

4-94

(a)

(e

)

(b)

(c

)

(d

)

Fig

ure

4.24

: Mic

rost

ruct

ural

det

ails

alo

ng

the

bond

ing

line

of a

sim

ilar

AZ

91D

-T6/

AZ

91D

-T6

wel

d. D

efin

ed b

ondi

ng li

ne in

the

uppe

r re

gion

s (a

) an

d (e

) w

ith a

very

ref

ined

tran

sitio

n zo

ne in

the

low

er s

tud

area

s (b

), (

c) a

nd (

d).

4-95

Figure 4.25: Second phase particles in regions highly plasticised within the welding zone.

(a) (b)

Figure 4.26: SEM micrographs taken over the BL with points for EDS analysis. In (a) particles stacked

along the BL and in (b) intermetallic compounds concentrated along grain boundaries.

Table 4.15: Results of the EDS analysis over the bonding line in similar AZAZ welds (wt.%).

Point Mg Al Zn Mn Si

1 2.23 39.64 --- 57.45 0.68

2 1.37 39.39 --- 58.71 0.56

3 60.86 32.91 3.88 2.36 ---

4 66.56 29.21 4.23 --- ---

5 67.14 28.74 4.12 --- ---

4-96

4.3.3 Mechanical Characterisation

4.3.3.1 Hardness Tests

As presented in Section 3.5.3, several microhardness profiles were obtained

vertically and horizontally along the welded area. Figure 4.27(a) to (f) show horizontal

profiles at 2 mm and 18 mm from the top base plate surface for different weld

combinations. In AEAZ joints, the welding process seems to have a positive influence

on the resultant microstructure of the consumable. While hardness values were

shown to vary between 40 HV and 50 HV in the BM, in the extruded area these

values range from 50 HV to 60 HV and from 60 HV to 70 HV in the upper and lower

weld areas respectively (see Figure 4.27(a) and (b). However, hardness values in the

RZ were practically unaltered with an increase in upsetting. On the AZ side of the

weld, one can note an increase in the hardness in comparison to the RZ, which to

some degree agrees with values found for the BM.

For MRI joints, material hardening in comparison with BM is easily observed from

start to finish of the extruded zone, resulting from the weld thermal cycle. Hardness

of the BM was shown to range between 46 HV and 54 HV, whereas in the extruded

area values varied generally between 60 HV and 70 HV in the upper and 60 HV and

80 HV in the lower weld areas (see Figure 4.27(c) and (d). Scans performed near the

top surface are very smooth and present a significantly lower scattering, even in

regions of the HAZ far away from the joint line on the AZ side of the weld. However,

in the lower stud areas, the 20 mm upsetting weld presented slightly superior

hardness values.

According to Figure 4.27(e) and (f), in AZAZ joints the hardness profiles also indicate

no apparent loss of hardness along the welding area. For 20 mm upsetting welds, the

hardness values varied randomly around 70 HV in the upper and between 75 HV and

80 HV in the lower weld areas. For the 10 mm upsetting welds, a significant increase

in hardness was observed in the lower region relative to the upper. Upper surveys

indicated values between 60 HV and 70 HV, while in lower scans values ranging

from 80 HV to 90 HV were obtained. Similar to the previous analyses, hardness

scattering in the upper survey was clearly reduced in relation to the lower

measurements.

4-97

(a)

(b)

(c)

4-98

(d)

(e)

(f)

Figure 4.27: Horizontal hardness profiles along the joints at 2 mm and 18 mm from the top surface.

4-99

4.3.3.2 Pull-out Tests

Similar to the results obtained in Section 4.2.4.2 for similar AZAZ-6/12 joints, AEAZ,

MRI and AZAZ welds within these welding parameters also always collapsed in the

stud material outside the deposit area. Since the failures occurred in the stud

material, very similar curves for 10 mm and 20 mm upsetting welds could be

observed, as demonstrated in Figure 4.28. As may be observed, up to approximately

10 kN all the curves are almost coincident. After this load is achieved, yielding

begins. Pull-out strength values were observed to vary from 25 kN to 30 kN for AEAZ

and from 30 kN to 35 kN for AZAZ joints. MRIAZ welds presented similar results,

indicating a maximum of 20 kN independent of the upsetting used.

(a)

(b)

Figure 4.28: Pull-out test results for 8 mm/10 mm upsetting welds.

4-100

4.3.3.3 Transversal Tensile Tests

Tensile properties were determined at room temperature for the base materials and

for the different welding combinations/parameters. Following the same procedure

presented in Section 3.5.5, two tensile specimens were obtained from each welded

sample, with the weld area placed in the middle of the gauge length. Tensile

strengths for dissimilar AEAZ welds are presented in Table 4.16 and summarised in

Figure 4.29. According to these results, AEAZ welds presented high strengths even

comparable with those of the AZ91D-T6 BM. For 20 mm upsetting samples,

specimens from the upper weld areas presented the higher average strength values

(166 MPa) in relation to those machined from lower areas (150.5 MPa). The

converse occurred for 10 mm upsetting samples, where the lower areas showed

higher average strengths (166.1 MPa) compared with the upper weld areas

(151.7 MPa). In terms of elongation, values ranging from 0.8 MPa to 1.8 MPa and

2.7 MPa to 1.6 MPa were observed for lower and upper samples in 10 mm and

20 mm upsetting welds respectively.

Table 4.16: AEAZ - Tensile testing results.

Sample Rm [MPa] Rp0,2 [MPa] A [%]

Upper 151 ± 14 123 ± 6 0.8 ± 0.5 AEAZ-10 mm

Lower 166 ± 8 118 ± 6 1.8 ± 0.3

Upper 166 ± 6 103 ± 3 2.7 ± 1.6 AEAZ-20 mm

Lower 150 ± 3 105 ± 4 1.6 ± 0.1

AE42 Base Material 104 ± 8 90 ± 2 4.5 ± 1

AZ91D-T6 Base Material 171.5 ± 16 140 ± 4 1.3 ± 0.7

Figure 4.29: Tensile test curves for 8 mm/10 mm upsetting welds in association with BM values.

4-101

MRIAZ welds also presented satisfactory strengths. As demonstrated in Table 4.17,

average values in the upper and lower weld areas respectively of 173 MPa and

150 MPa were obtained for MRIAZ-10 mm welds, whereas slightly lower values of

159 MPa and 147 MPa were found for MRIAZ-20 mm joints. Elongations were

observed also to match the BM values and varied between 1.0 % and 1.6 % and

between 0.8 % and 2.0 % respectively for 10 mm and 20 mm upsetting welds in the

lower and upper weld areas.

Table 4.17: MRIAZ - Tensile testing results.


Upper 173 ± 8 129 ± 5 1.6 ± 0.1 MRIAZ-10 mm

Lower 150 ± 2 124 ± 2 1.0 ± 0.1

Upper 159 ± 4 103 ± 6 2.0 ± 0.3 MRIAZ-20 mm

Lower 147 ± 9 123 ± 4 0.8 ± 0.1

MRI230D Base Material 94 ± 8 90.1 ± 2 0.8 ± 0.2


According to Table 4.18, AZAZ-08/10-20 welds presented values that to some extent

were similar to those of the BM. Specimens welded with 20 mm upsetting present

average strength values of 184 MPa and 159 MPa respectively for upper and lower

stud areas. Following the same tendency, tensile strengths of 10 mm upsetting welds

showed average values varying from 156 MPa to 175 MPa respectively for lower and

upper samples. On the other hand, yield strengths seem to differ widely from BM

values, particularly in the upper weld areas. Values around 116 MPa and 118 MPa

were obtained for 10 mm and 20 mm upsetting welds respectively.

Table 4.18: AZAZ - Tensile testing results.


Upper 175 ± 3 116 ± 4 2.1 ± 0.3 AZAZ-10 mm

Lower 156 ± 7 141 ± 7 1.0 ± 0.5

Upper 184 ± 3 118 ± 4 2.1 ± 0.1 AZAZ-20 mm

Lower 159 ± 6 123 ± 13 1.2 ± 0.5


4-102

4.4 JOINT PERFORMANCE

4.4.1 Creep Tests

4.4.1.1 AE42 Studs (Reinforcement)

Creep tests conducted in AEAZ welds, according to the procedures presented in

Section 3.5.7, presented a very similar behaviour between lower and upper samples

(see Figure 4.30). At 125 °C the extruded material (lower sample) demonstrated a

slightly better performance, while at 150 °C the upper (ref.) sample showed a

superior creep resistance after 100 hours. Creep compression indicated values of

1.42 % and 1.79 % respectively for lower (weld) and upper (ref.) samples at 125 °C

after 100 hours. At 150 °C compression values of 4.1 % and 5 % for upper and lower

samples respectively were observed. In this case, the reference sample presented a

higher deformation within the first stage of the test. However, after approximately

65 hours the strain observed in the reference sample was surpassed by that of the

lower sample, which demonstrated a higher creep rate in the secondary region. At

175 °C the AE42 material suffers an expressive creep strain in both cases.

Figure 4.30: Creep curves of dissimilar AEAZ welds in compression at 125 °C, 150 °C and 175 °C with

80 MPa.

4-103

4.4.1.2 MRI230D Studs (Reinforcement)

As expected, MRIAZ welds presented a very high performance among the tested

combinations. Lower and upper samples demonstrated very similar creep behaviour

at lower temperatures (see Figure 4.31) with creep compression values around 0.3 %

after 100 hours. At 150 °C, the reference sample creeps at a higher rate during the

first stage being, however, overtaken by the lower sample somewhere in the middle

of the test, as observed for AE42 Studs. After 100 hours, creep compression values

of 1.4 % and 2 % were obtained for upper and lower samples respectively. At 175 °C

the upper reference sample creeps around 3 %, while the lower sample creeps

between 9 % and 10 % after 100 hours at 80 MPa.

Figure 4.31: Creep curves of dissimilar MRIAZ welds in compression at 125 °C, 150 °C and 175 °C

with 80 MPa.

4.4.1.3 AZ91D-T6 Studs (Reinforcement)

According to Figure 4.32, in similar AZAZ welds, the samples extracted from the

lower stud areas presented in all three cases a reduced creep resistance compared

with the upper samples (ref.). Primary creep regions are relatively short and exhibit a

sharp transition to steady state creep in tests performed at lower temperatures.

Under these temperatures, creep compression around 0.6 % and 1.3 % were

observed after 100 hours for upper and lower samples respectively. However, a clear

4-104

transition from stage II to III was not observed, independent of the temperature. At

125 °C and 150 °C (only for the reference sample) an evident steady creep stage,

characterised by the minimum creep rate extending to the end of the experiment,

was observed. At 150 °C the reference sample presented 2 % creep compression,

while the lower one experienced values around 10 % after 83 hours. At 175 °C the

unsuitability of AZ alloys for high temperature applications became clear, with creep

compression values of 10 % after 10 and 14 hours for lower and upper samples

respectively.

Figure 4.32: Creep curves of similar AZAZ welds in compression at 125 °C, 150 °C and 175 °C with

80 MPa.

4.4.2 Bolt Load Retention Tests

As presented in the experimental procedure (see Section 3.5.6), bolt load retention

(BLR) tests were performed for the different configurations. In the first case, as

presented in Figure 3.12, welded joints would be subjected to tensile stresses,

whereas in the second case the reinforced area would be subjected to retention tests

in compression (see Figure 3.13). Both nomenclatures, i.e. BLR in tension and BLR

in compression, are used only within this study and make reference to the position of

the reinforced component in relation to the couple.

4-105

4.4.2.1 BLR Tests – In Tension

Bolt load retention tests (in tension) within this study were performed in a variety of

samples welded with different parameters and subjected to different testing

conditions. In order to keep the scope of this work within reasonable limits, only the

most relevant results are reported in this section.

Figures 4.33 and 4.34 show the remaining load in tests carried out at 175 °C after

100 hours. Figure 4.33 shows AZ91D-T6/BM bolted with steel screws having after

100 hours at 175 °C remaining load values of around 30 % and 1 % of the preload

force for joints tightened with 40 Nm and 60 Nm respectively. In comparison, under

the same circumstances the MRI230D base material presented retaining loads of

75 % and 35 % in relation to the initial preload force for respectively 40 Nm and

60 Nm tightening torque. BLR in reinforced joints, in this context, presented an

intermediate behaviour between AZ91D and MRI230D. Joints produced with larger

diameter studs show a better performance than those produced with 6 mm studs.

After 100 hours, welds produced with 6 mm studs retained 36 % and 11 % of the

initial load for joints tightened with 40 Nm and 60 Nm respectively. For 8 mm stud

welds, however, higher values of around 41 % and 20 % were observed under the

same conditions. It was also clear to see that higher tightening torques lead the

couple to inferior remaining loads. Joints tightened with 60 Nm retained 11 % and

20 % for 6 mm and 8 mm geometries respectively, while 40 Nm screwed couples

presented 36 % and 41 % remaining loads, also for 6 mm and 8 mm geometries.

Furthermore, welds produced with 8 mm studs retained higher loads (41 % and 20 %

for 40 Nm and 60 Nm tightening torques) than those retained by 6 mm stud welds,

which presented 36 % and 11 % for 40 Nm and 60 Nm tightening torques

respectively.

Welded joints tightened with aluminium fasteners (see Figure 4.34) presented clearly

higher remaining load values, varying from 63 % to 69 % of the initial preload force.

Also in this case, the load remaining in welded joints were higher than those

observed for the AZ91D-T6 BM, but lower than those verified for the MRI230D BM.

For 6 mm stud welds, an uncommon trend was observed, in which samples tightened

with 35 Nm showed a slightly higher remaining load (64 %) in relation to those

torqued with 25 Nm (63 %). However, as in the previous case, joints produced with

4-106

larger diameter studs showed a better performance than those produced with 6 mm

studs and welds produced with 8 mm studs retained higher loads (69 % and 66 % for

25 Nm and 35 Nm tightening torques) than those retained by 6 mm stud welds, which

presented 63 % and 64 % for 25 Nm and 35 Nm tightening torques respectively.

Figure 4.33: Remaining load in tensile stress retention tests at 175 °C for dissimilar MRIAZ joints with

steel fasteners. MRI230D and AZ91D-T6 base material results are also presented for comparison.

Figure 4.34: Remaining load in tensile stress retention tests at 175 °C for dissimilar MRIAZ joints with

aluminium fasteners. BM results are also presented for comparison.

4-107

4.4.2.2 BLR Tests – In Compression

In order to investigate BLR also in compression, welded samples were tested under

different conditions. In experiments carried out at 175 °C using steel bolts of strength

grade 8.8, Mg samples (AZ91D-BM, AEAZ, MRIAZ and AZAZ welds) were tightened

with a pretensioning force of 20 kN against a threaded Al AlSi9Cu3 specimen (see

Figure 3.13). Labelled points PI (= initial load at room temperature), Pk (= highest

load attained by the couple during heating at test temperature), PR (=retained load at

test temperature) and PF (= final load retained by the couple after returning to room

temperature) indicate the critical loads, which define the bolt load behaviour of the

investigated alloys. In terms of the overall BLR behaviour and its effect on engine

performance, the two most significant loads are the initial load PI and the load at the

completion of the test after returning to ambient temperature PF. The important ratio

is therefore PF/PI as the engine is cycled from room to operating temperature and

back again. The ratio PR/PK represents the fraction retained at test temperature,

whereas PF/PI the fraction retained after the joint returns to ambient temperature. The

overall relaxation behaviour (PI - PF) is made up of two major components: the

compressive creep component (PK - PR) and a component due to yielding, which is

obtained from the difference. These important factors are presented in the following

table for the different materials and joints under investigation.

Based on Figure 4.35 and Table 4.19, Mg AZ91D-T6 seems not to be suitable for

applications at 175 °C, even if reinforced with high-temperature resistant materials.

This is because the BLR test PF values, which gives the final load retained by the

couple after returning to room temperature, dropped below the zero level in all the

investigated cases. The ratio PF/PI presented values around zero, which means that

no load is retained by the couple after returning to room temperature. The similar

AZ91D-T6 welded joint followed the same behaviour revealed by creep tests and

clearly presented the lowest remaining load.

4-108

Figure 4.35: Typical bolt load compressive stress retention curves at 175 °C for AZ91D-T6 similar and

AEAZ, MRIAZ dissimilar welds covering the entire test. Base material behaviour is also presented for

comparison.

Table 4.19: Critical loads in (kN) for Mg welds tested at 175 °C.

Temp. Alloy PI PK PR PF PR / PK PF / PI PI - PF Creep Yield

AEAZ - Weld 20.6 26.9 8.06 -0.9 0.30 -0.04 21.50 18.84 2.66

MRIAZ - Weld 20.3 26.2 8.86 -0.5 0.34 -0.02 20.80 17.34 3.46

AZ91D - Weld 19.7 24.9 4.59 -1.15 0.18 -0.06 20.85 20.31 0.54 175 °C

AZ91D - BM 19.7 26.4 7.53 -0.2 0.29 -0.01 19.90 18.87 1.03

Since BLR tests at 175 °C demonstrated that AZ91D-T6 alloys, even reinforced,

were not suitable to be used under these conditions, tests at 150 °C were performed

in AZ91D-T6/BM, MRI230D/BM and MRIAZ welds with steel and Al bolts. Figure 4.36

illustrates the plots depicting bolt load retained after 50 hours at 150 °C for 15 kN and

10 kN. As expected, from Figure 4.36 and Table 4.20 one can clearly observe the

higher PF/PI ratio (the retained load after joint returns to room temperature) for the

MRI230D/BM, confirming the superior high temperature properties of this alloy.

MRI/BM sustained a PF value of 6.81 kN, which is approximately 44 % of the initial

load (PI), after being subjected to the temperature cycle. MRIAZ welds tested under

the same loading condition records a PF value of 6.49 kN, which is approximately

41 % of PI. AZ91D BM, on the other hand, retains only 24 % of the initial load. The

joint bolted with an Al fastener presented an untypically low remaining load of only

27 %, which was to some extent unexpected.

4-109

Figure 4.36: Typical bolt load compressive stress retention curves at 150 °C for MRIAZ dissimilar

welds with steel and aluminium bolts covering the entire test. MRI230D and AZ91D-T6 base materials’

behaviour is also presented for comparison.

Table 4.20: Critical loads in (kN) for Mg welds tested at 150 °C.

Temp. Alloy PI PK PR PF PR / PK PF / PI PI - PF Creep Yield

MRI230D/BM 15.4 22 18.4 6.81 0.84 0.44 8.59 3.60 4.99

AZ91D-T6/BM 15.6 22.1 13.3 3.68 0.60 0.24 11.92 8.80 3.12

MRIAZ - Weld Steel Bolt

16 23.9 16 6.49 0.67 0.41 9.51 7.90 1.61 150 °C

MRIAZ - Weld Aluminium Bolt

10.3 14.2 8.36 2.77 0.59 0.27 7.53 5.84 1.69

5-110

5 DISCUSSION OF RESULTS

5.1 BASE MATERIAL CHARACTERISATION

5.1.1 Mg-Al-Zn [AZ] – Ingot (heat treated to T6 condition)

According to the EDS results listed in Table 4.2, round particles within the Mg-matrix

corresponding to points 1 and 7 to 11 showed a high concentration of Al and Mn with

a minor presence of Si. Such compounds are suggested to be AlMn based phases,

commonly present in AZ alloys. Points 2 to 4 and 6 indicate a high concentration of

Mg and Al with some significant evidence of Zn. Such Mg containing phases (Mg-(Al-

Zn)) have been reported in the literature [14] to be present in Mg-Al-Zn alloys and are

suggested to be Mg17(Al, Zn)12 second phase products. Point 5 reveals the average

composition of the Mg-matrix. No clear presence of Mg2Si particles was observed

within the investigated points, although its presence has also been reported in AZ Mg

alloys. Since the lamellae structure covered most of the material’s macrostructure, to

count the average grain diameter was difficult. However, microscopic observations

with higher magnification revealed the traces of grain boundaries delineated by

undissolved particles.

Table 4.1 shows the AZ91D chemical composition, which contains no drastic

differences when compared with values stated in the literature. However, although

the AZ91 ingots were heat treated to homogenise the microstructure, mechanical

testing also indicates significant differences between properties in the lower and the

upper part of the ingot. Furthermore, values ranging from 50 % to 75 % of the

strength stated in the literature [14,16,156] were obtained during the mechanical

characterisation.

5.1.2 Mg-Al-Rare Earths [AE] – Ingot (cast condition)

AE42 cast alloys, under die cast conditions, are shown in the literature [27,150,151,157] to

form an eutectic phase near the grain boundaries. According to Figure 4.3 and in

agreement with prior investigations [29], SEM analysis revealed that grain boundaries

of AE42 base material present widespread precipitation of second phase particles

5-111

with lamellar morphology across the interdendritic microstructure. In Table 4.5, points

1 and 2 reveal the average composition of the Mg-matrix. Although the chemical

analysis indicates 4 % Al in weight, the matrix has rarely over 2 % Al. This

characteristic may be explained by the fact that the interdendritic phases are mostly

composed by Al-based compounds. Furthermore, the Al content in solid solution is

somehow limited, firstly because of the stability of the Al-RE phases and secondly

due to the limited solubility of Al in Mg at room temperature. Points 3 and 4 indicate a

high concentration of Al and Mn with some significant amount of RE, indicating

possibly the presence of the classical Mn-(Al-RE) phase formation. Such Mn-

containing phases (Mn-(Al-RE)) have been reported in the literature [112] to be present

in minor amounts in the “as-cast” condition and are suggested to be Al10RE2Mn7

particles. Scans corresponding to points 5 and 6 showed a high content of Al and RE

with minor traces of Mn. Such compounds are suggested to be Al-RE based phases,

possibly Al11La3, Al4La or Al2Nd, commonly present in AE Mg alloys. Evidence of Si

has been observed, but considered an impurity, although it is known that Mg2Si can

be formed in some other Mg-alloys.

Mechanical characterisation of the AE42 BM, as demonstrated in Table 4.6 and

Figure 4.4, has pointed out a significant difference between the lower and the upper

part of the ingot in the longitudinal direction. Furthermore, mechanical properties

obtained during the course of this study differ from those available in the literature [8,14,16]. In the transverse direction, however, tensile strength results presented a large

scattering, independent of their position in relation to the ingot.

5.1.3 Mg-Al-Ca-Sr [AXJ (MRI)] – Ingot (cast condition)

Similar to the AE42 alloy, chemical composition and mechanical properties were

analysed and indicated significant differences between the obtained values and

those found on the literature [33]. According to Table 4.7, a significant Sn content was

identified in the actual composition, which is in disagreement with the literature.

Furthermore, tensile strength values, shown in Table 4.9 and Figure 4.6, were found

to strongly vary, not only in relation to the literature, but also between the lower and

the upper portion of the ingot. Strength values (i.e., Rp and Rm) were significantly

lower than those indicated in the literature [8,33], as expected for an ingot, which was

chilled cast.

5-112

Additions of Ca and Sr in the MRI230D alloys increase their creep resistance, due to

the formation of stable Al-based precipitates instead of the detrimental Mg17Al12 in the

grain boundaries [33]. The rest of free Al is reported to form Al-Ca precipitates

(specifically Al2Ca). EDS analysis (see Figure 4.5 and Table 4.8), corresponding to

points 1 to 6, showed at grain boundaries intermetallic compounds with compositions

varying in a broad range. Points 1 and 5 indicate a high concentration of Al, Mn and

Zn with some significant traces of Sn. Such compounds are suggested to be former

AlMn-based phases, commonly present in AZ alloys that apparently incorporated

some Zn and Sn during the casting process. According to the chemical analysis (see

Table 4.7) Zn was found in very low percentages while Sn, although it does not

compose the alloy [33], was present in a high content. Points 2 and 3 evidence a

substantially high Sn concentration with a minor presence of Mg and Ca, whereas

points 4 and 6 indicate the fine distributed Mg-Al-Ca precipitates with a large

variation in chemical composition. From this analysis, it is evident that the formation

of eutectic Mg17Al12 phase was completely suppressed in the presence of Ca and Zn,

as reported in the literature [43]. Points 7, 8 and 9 reveal the average composition of

the α-matrix, which seems to be very homogeneous. No clear presence of Al4Sr was

identified in the investigated points, although its presence has been reported in MRI

Mg alloys [40]. Round particles within the Mg-matrix, corresponding to points 10 to 13,

showed a similar concentration as revealed by points 1 and 5. However, in this last

case, the particles are randomly distributed inside the α-Mg matrix and not grouped

along the grain boundaries, as observed in the previous case.

Summarising the base material characterisation generally, significant differences

were observed concerning not only chemical composition, but also mechanical

properties in the selected BMs. In relation to chemical composition, the presence of

Sn and some indication of Zn in the EDS analysis of MRI230D alloys can be

considered the most prominent changes in comparison with values found in the

literature [33]. Additions of Sn serve to increase ductility, to reduce the tendency of

cracking while being hot worked and increase high temperature properties due to the

high temperature stability of the Mg2Sn phase formed [14,23]. On the other hand, Zn is

often used in combination with Al to produce an improvement in the room

temperature strength of Mg alloys. A further evaluation or discussion involving this

material is difficult draw, since not very much information on this alloy is available in

5-113

the literature. Furthermore, a significant difference concerning mechanical properties

was observed, not only between the lower and the upper part of the ingots, but also

between the values obtained and those reported in the literature. A sensible reason

for such differences could be attributed to the fact that typical mechanical properties

and characteristics found in the literature are normally assessed using thin cast

plates. In those cases, a refined microstructure is formed on the outer regions (skin)

of the plate, due to the faster cooling gradients imposed at these locations. This

refined microstructure contributes to enhance the mechanical properties, therefore

achieving the high strength values reported in the literature. However, in this

investigation tensile samples and later studs were extracted from the middle of the

ingots avoiding the beneficial influence of the skin microstructure and consequently

generating lower strengths. Additionally, chilled casting rather than die casting

samples were used in this work, which also contributes to the inferior strengths

observed during mechanical testing. Finally, the fact that the specimens machined

from positions 1 to 5 (see Figure 4.2) indicated higher strengths than those from 6 to

10 can be directly related to the casting process. Regions 1 to 5 are in contact with

the mould, and so experience higher cooling rates and probably present a refined

microstructure and consequently generating samples with higher strengths.

5.2 SIMILAR AZ91D-T6 JOINTS – 6 MM DIAMETER/12 MM UPSETTING

5.2.1 Process and Thermal History

Based on the results presented in Figure 4.8, the equipment showed a sufficient

repeatability with a relatively low standard deviation within the investigated range of

parameters. One can also observe a considerable decrease in the standard deviation

values as soon as welds with higher pressures are performed, i.e. towards the middle

of the control range of the system. However, in some particular cases welds

performed with the same parameters showed a large variation in welding time (see

Figure 4.8(a) – welds 1C and 1D). Since the welding system has been assessed in

previous studies and demonstrated to have a high repeatability, the possible source,

which could cause such deviations on the welding cycles in welds carried out under

the same conditions, was directed to the material itself. As presented in section 4.1.1,

the upper and lower side of the ingots have significant differences with regards to

their mechanical properties. Furthermore, as a high Al content gravity casting alloy,

5-114

AZ91D may present some regions with grouped pores, which maybe was the cause

for the early collapse observed in some samples during mechanical characterisation

(see Figure 4.2). Considering the present study, studs from different parts of the

ingots, machined in different directions (longitudinal and transversal, depending on

the dimensions of the ingots) were welded within the same group. In this context, a

region with fine grouped pores could have been present inside the stud, which would

influence the upsetting rate during the welding (see as an example Figure 4.8,

specimen 2B). As soon as this region achieves the frictional interface during the

welding, the upsetting rate increases instantaneously, causing interference to the

welding progress. Both intrinsic characteristics of the material have to be considered

in a future approach and these have possibly influenced and contributed to increase

the dispersion of the process stability results.

Figure 4.8 also highlights a significant influence of the welding pressure on the

welding time. By duplicating the pressure, the welding time is reduced by more than

60 %, from 24.4 to 9.5 seconds on average. Moreover, although the motor is

electronically stopped by proportional valves, the total upsetting (TU) was observed

to be significantly higher than the preset value. As observed in Figure 4.8, the

upsetting during the stopping time (see UST values), which is a non-controlled event,

increases significantly the TU. Generally, a higher forging force tends to increase the

axial shortening during the stopping time (UST) leading the remaining length of the

welded pair to be substantially smaller. Figure 4.8 indicates on average a UST value

of 2.4 mm for low pressure welds, while 3.3 mm upsetting values were observed for

higher pressure welds. In this context, if the welding system is upsetting controlled,

the welding parameters should be accurately monitored and controlled if the final

length of components plays an important role.

The accepted opinion is that the use of a forging force significantly improves the

mechanical properties of the weld [158] by breaking up the coarse inclusions, which

are adversely reoriented during frictional heating, and by refining the coarse grains

by hot-working [47,60]. However, some experiments show that good welds can be

obtained without an increase of the force after rotation stops, which means, without

any forging force [61].

5-115

Both tendencies discussed above are summarised in Figures 5.1 and 5.2. Figure 5.1

shows the UR dependency on the welding pressure, while the variation of the

upsetting during the stopping time (UST) with the forging pressure is presented in

Figure 5.2.

Figure 5.1: Upsetting rate (UR) dependency as a function of welding pressure.

Figure 5.2: Variation of the upsetting during the stopping time (UST) in relation to the forging pressure.

According to Figure 3.4 (see Section 3.2), temperature monitoring was performed

with two thermocouples along the bonding interface with the upwards moving

frictional plane considered the heat source. It should be also emphasised that with

5-116

this experimental set-up only the temperature within the heat affected zone (HAZ) on

the sides of the holes were measured. Additionally, it has been observed that, owing

to the difficult procedure during the spark erosion of the thermocouple holes, the

actual distance between the tip of the holes and the bonding line varied significantly,

although 500 μm was specified. Although Mg materials have a high thermal

conductivity, which could compensate for small variations in thermocouple hole

depth, these variations complicate the analysis and the direct comparison of results.

Hence, the measurements presented should only be considered as general

indications and not as an absolute thermal analysis.

In all the records, the maximum temperature was reached at the end of the weld, as

demonstrated in Figure 5.3. Some milliseconds after the stud has touched the cavity

bottom, the lower thermocouple (T2) starts to record the temperature. Once the

process reaches a steady state, i.e. the temperature and height of the plasticised

zone is kept in equilibrium, the frictional interface starts to move upwards.

Temperature indication T2 increases rapidly within the first seconds due to its

proximity to the heat source. As soon as the frictional interface reaches a position

parallel to thermocouple T2 (see Figure 5.3) the first peak temperature is observed,

since the heat source has its closest position in relation to T2. As the process

continues, the frictional interface moves further upwards, causing a drop in the T2

measurements due to conduction losses on the way between the heat source and

the thermocouple. However, with increasing thermal saturation of the base plate, at a

certain point temperature T2 reaches a minimum and starts to increase again up to

the end of the welding process. On the other hand in T1, positioned 6 mm from the

top surface, a slight rise in the temperature field is observed after the first steep

increase at the beginning of the process, due to the increasing thermal saturation of

the base plate. Contrary to T2, T1 does not show any peak during the welding

progress. This phenomenon can be explained by the fact that when the frictional

interface (in the form of an elliptic paraboloid) achieves T1 level, the process reaches

the limiting control parameter, i.e. the last frictional interface coincides with the

position where the upper thermocouple is placed. At this stage, the maximum

temperature is achieved after the deposition process has finished.

5-117

Figure 5.3: Welding progress diagram combined with temperature measurements (welding pressure

6.90 bar, forging pressure 20.7 bar, rotational speed 4000 rpm and 12 mm upsetting).

Bearing in mind the different frictional interfaces at the beginning and at later stages

of the process (see Figure 5.4), the different temperatures measured along the hole

depth (as demonstrated in Figure 5.3) can be better understood. The temperature

development changes due to the different heat generation and due to the different

cooling conditions at initial and final welding stages. On the one hand the frictional

interface, i.e. the area between the stud tip and the deposited material in direct

contact and relative movement (see Figure 5.4), changes from the basic stud

diameter at the beginning of the weld (see Figure 5.4(a)) to an elliptic-paraboloid

shaped surface in later stages (see Figure 5.4(b)). Owing to the increased contact

area, since the stud is tapered, the heat production is higher, thereby generating the

higher temperatures at the final stages of the weld. On the other hand before the

touch-down distance is achieved, the base plate is situated in air at room

temperature, which explains the steep temperature increase at initial stages of

around 130 °C/s. As the process advances, a thermal saturation seems to occur in

the base plate, leading the temperature to show a reluctant increase during the final

stages. From this point (see grey dashed arrows indications in Figure 5.3) the

temperature experiences an increase of around 2.28 °C/s up to the final stage of the

process.

5-118

(a) Ups: 4 mm. (b) Ups: 10 mm.

Figure 5.4: Frictional interface after 4 mm (a) and 10 mm (b) upsetting.

While welding times showed a variation of more than 60 % by duplicating the

pressure, peak temperatures presented a variation of around 4 % and 9 % for T1 and

T2 respectively under similar conditions, i.e. by duplicating the pressure (see Figure

4.9). Moreover, although the peak temperature is not substantially affected, the dwell

time (time at maximum temperature) and cooling rate are substantially affected. As

an example, for low pressure welds the temperature remains for more than 20 s over

400 °C in some cases in the upper weld areas (see Figure 4.9(a)). On the other

hand, for higher pressure welds, the time at peak temperature decreases and in

some cases does not even achieve the same level (see Figure 4.9(e)). By

approximately equal average peak temperatures, for some alloys the dwell time and

cooling rate might be of substantial importance for precipitation/transformation

phenomena. Moreover, the extension of the heat affected zone can be directly

related to the temperature curves presented in Figure 4.9.

The mean values of peak temperatures (see Figure 4.9) revealed that temperatures

very close to the melting point were achieved. Figure 2.2 shows the Al-Mg phase

diagram, which points out a solidus temperature of approximately 440 °C for Al-Mg

based alloys. Considering that the measurements were carried out some

micrometers away from the bonding line, the solidus temperature was perhaps

overcome in regions very close to those rubbing areas. As determined by Bowden et

al. [159,160] there is an extremely steep temperature gradient close to the frictional

interface, i.e. the temperature decreases remarkably over an extremely small

distance from the interface. This statement supports the theory of achieving

5-119

temperatures above the solidus line in regions close to the frictional interface.

Experimental evidence supporting local melting during friction based processes in

Mg-alloys has been provided by a number of investigators [161-163]. As reported in the

literature [164,165], once the solidus temperature has been overcome, local melting

occurs and the viscosity of plasticised material decreases, so that the rate of heat

generation suddenly decreases. However, when the material’s temperature falls, the

local molten films at the interface solidify, causing their viscosity and consequently

the heat generation rate and temperature to increase. This process occurs

repeatedly, so that the peak temperature at the bonding line (BL) attained during

friction welding is the solidus temperature of the materials involved [166]. Figure 5.5

shows a micrograph taken over the BL, indicating that low melting point intermetallic

eutetics, concentrated on the grain boundaries, could be segregated and molten

during the welding process. Furthermore, the micrograph in Figure 5.5 seems to

show epitaxial grain growth, which would suggest (or even confirm) the presence of

incipient melting at the stud/BM interface. Considering local changes and dissolution

of the second phase particles (see Section 4.2.3), the microstructure around the BL

is very different from that observed in the cast material. Many publications in the

literature [164-166] support the theory that microstructural features in friction welds are

determined by a combination of grain boundary sliding and the limitation of cavity

interlinkages when transient local molten films are formed, based on the relationships

between strain rate/resultant microstructure.

Figure 5.5: SEM micrography taken from the welding zone with indications of transient local melting.

5-120

In terms of heat input, Vill [50] proposed that the heat generation N during friction

welding can be determined based on the following relation:

[kW] [Equation 5-1]

Where,

p – Axial pressure [kg/mm2] K – Material dependent constant [mm2/min2]

n – Rotational Speed [1/min] r – Radius of the cross section [mm]

Considering Equation 5-1 and extrapolating its use for FHPP, one can observe that

heat generation increases proportionally with welding pressure. Since the parameters

n, K and r were constant, welds from group 3 would be expected to generate the

higher amount of energy per time. In fact, Figure 5.9 showed that with higher axial

pressure (group AZAZ-3) the temperature rises with about 180 °C/s, while lower axial

pressures result in a heating rate of 120 °C/s (group AZAZ-2) and 80 °C/s (group

AZAZ-1). However, the global heat generation is time-dependent and therefore

strongly influenced by the welding time. Table 5.1 lists the heat generation per time

and the consequent global heat generation. Comparing samples “C”, selected to

further metallurgical analysis, it can be seen that lower pressures (sample AZAZ-1C)

have generated the higher amount of heat. However, sample AZAZ-3C produced

approximately 65 % of the heat generated under lower pressures.

Table 5.1: Global heat generation for AZAZ welds.

Sample Welding Pressure

Heat Generation N (1) [W]

Welding Time [s]

Global Heat Generation N’ (2)

[J]

1C 6.90 bar (1.0) 1.0*(2Kr/n) 29.5 29.5*(2Kr/n)

2C 10.34 bar (1.5) 1.5*(2Kr/n) 17 25.5*(2Kr/n)

3C 13.79 bar (2.0) 2.0*(2Kr/n) 9.9 19.8*(2Kr/n)

(1) Heat generation N (in J/s), considering the welding pressure; (2) Global heat generation N’ (in J), considering the welding pressure and the welding time.

5-121

5.2.2 Metallurgical Characterisation / Joint Formation Mechanisms

Metallurgical characterisation, as presented in Section 4.2.3, shows significant

microstructural changes in the lower weld areas, resulting from different welding

parameters. Considering that temperatures at the bonding line may have achieved

even higher values than the maximum indications pointed out by thermocouples T1

and T2 (respectively 448 °C and 395 °C, 383 °C and 281 °C for samples 1C and 3C,

see Figure 4.9), it seems that the axial pressure definitely plays an important role in

the final structure around the welding area. Not only the different dwell times, but also

the different cooling rates (see Figure 4.9) imposed by the process influence the joint

formation in a decisive manner. Such different thermal cycles possibly allow

solubilisation, precipitation and diffusion mechanisms to take place, establishing the

changes in the microstructure after welding. However, to some extent the average

grain size inside the extruded material remains similar for samples welded under

different conditions. In this case, it seems that the different welding times and their

consequent different cooling times (see Figure 4.9) have a minor influence and

therefore do not change the microstructure drastically in this particular region.

Furthermore, lower weld areas indicated smaller grain sizes compared to the upper

part. The higher temperatures measured in the upper weld areas seem to be a

reasonable cause of the unequal structure along the weld depth. The larger contact

area in the upper regions, with a correspondingly higher heat generation, causes a

greater volume of material to be heated, leading to wider affected regions.

Welds performed at lower pressures tend to result in a joint with a very

homogeneous microstructure, associated with a smooth deposition process. On the

other hand higher axial pressure tends to cause a more irregular deformation of the

stud and consequently a very inhomogeneous microstructure along the hole’s depth.

Figure 5.6 shows the macrograph of the weld AZAZ-3C, where some areas with non-

plasticised stud material were observed. Points “A”, “B” and “C” indicate points where

the stud shears off during the deposition process (welding).

5-122

Figure 5.6: Macrograph of the AZAZ-3C joint.

Since the weld progress was monitored by a data acquisition system, based on that

the mechanism of joint formation and its particularities can be outlined for each

specific case. In joint 3C, presented in Figure 5.6, as soon as the stud touches the

bottom of the cavity, heat is produced, generating a local plasticised layer in this

region. Due to the very high axial pressure, the plasticised material develops at a rate

slower then the axial feed of the consumable rod. In turn, it means that the frictional

rubbing surface (frictional interface) does not have, under these conditions, enough

time to rise along the consumable rod to form the dynamically recrystallised deposit

material. Instead of being gradually consumed, lower stud areas rub against the

cavity bottom and are, due to the heat generated, deformed and assume the positive

form of the hole. With the increasing contact area, the torque increases

correspondingly, up to a point where the frictional moment becomes larger than the

stud’s internal shear resistance. At this point, indicated as point “A” in Figure 5.6, the

stud shears off and starts a new frictional plane on the top of this deposited material.

The shearing-off of the stud can be also followed on the DAS measurements (see

Figure 5.7). After the stud has sheared-off the frictional torque reduces abruptly, as a

result of the significantly smaller frictional interface. This smaller frictional interface

causes, in turn, a sudden increase on the rotational speed. After some milliseconds,

the monitoring system brings the motor again to the preset rotational speed. Further

5-123

analysis can also be made in terms of temperature progress (see Figure 5.7). After

the process has been initiated, the temperature T2 increases steeply. However, after

the first stud shearing-off phenomenon, the heat generation reduces accordingly, as

a consequence of the smaller frictional interface. The temperature drops for a while

and starts to rise again at a lower rate as the process continues. Two further

shearing phenomena can be identified in Figure 5.7, in agreement with the

macrograph shown in Figure 5.6. At those locations the process “jumps” over a

certain stud distance without the stud material being plasticised in between (see

Figure 5.6 between indications “B” and “C”) and starts again with a new frictional

plane on top of this sheared material. This irregular microstructure can lead the joints

to an unexpected collapse and should therefore be avoided by means of optimal

process parameters. The main outcome of Figure 5.7 can be related to the

monitoring of the process. In other words, a sequence of welds could be monitored

and, after a statistical treatment of welding data, an optimal range could be

established where homogeneous welds are produced. Such control would be

meaningful for mass production and could be used as on-line monitoring for quality

control purposes.

Figure 5.7: Welding parameter measurements of weld AZAZ-3C. The shearing phenomena observed

on the macrograph could be clearly identified.

5-124

The results from the EDS analyses within the recrystallised zone, as presented in

Table 4.11, suggest some changes in comparison with the cast microstructure. The

grain boundaries in the extruded material within the RZ, corresponding to points 1 to

4 (see Figure 4.14), show in the first case a high concentration of Mg and Al with a

minor presence of Zn. Despite the unfavourable stoichiometry shown by the EDS,

such an analysis possibly indicates the formation of Mg17Al12 compounds around the

primary growing Mg-Al solid solution, as a result of a non-equilibrium local

solidification under rapid freezing conditions. Points 5 to 7 indicate an exceptionally

high concentration of Al in the α-Mg phase inside the fine recrystallised grains. It

seems that the absence of Al observed in the formation of the grain boundary

structure was partially transferred and dissolved into the matrix, thereby elevating the

Al content in those regions. Compared with EDS in the base material distant from the

weld, points 8 to 10, one can clearly observe the enrichment of the α-Mg phase with

Al in the above mentioned regions. Such points show the real composition of the

matrix with around 7 % Al. Although the chemical analysis indicates around 9 % Al by

weight, the matrix has rarely over 7 % Al. This characteristic may be explained by the

fact that the interdendritic phases are mostly composed of Al-based compounds, i.e.

the Al content in solid solution is somehow limited, due to the restricted solubility of Al

in Mg at room temperature, which is compensated by the formation of high Al content

second phase particles.

In order to monitor the joint formation, a sequence of welds were produced with an

increasing upsetting from 2 mm to 12 mm (the full upsetting for such welds) in 2 mm

increments. After the set upsetting was achieved the weld was interrupted and the

rotating consumable was pulled out of the cavity. Figure 5.8 shows the gradual filling

of the hole with increasing upsetting. Bearing in mind that in friction welding heat is

produced due to friction, which transforms the kinetic energy into heat, as soon as

both faying surfaces are in contact, heat begins to be produced. At the very first

process stages the heat produced is only used to smooth both surfaces and to

plastically deform the consumable. Due to the high heat generation, the consumable

becomes the positive form of the cavity up to a certain length (see scratches in the

sidewall in the white detail in Figure 5.8(a)) before it starts to be deposited. This initial

phase is characterised by seizing and microbonding processes, where the surfaces

reach locally an extremely high temperature, promoting plastic deformation and

5-125

hence flattening the faying surfaces, particularly in the outer regions where the

surfaces are subjected to the highest relative rotational speed. Isolated local bonding

that eventually takes place across the contact surface is almost instantaneously

sheared off. The shearing of those microbonds induces additional heat in the

surrounding material, contributing to an increase in the instantaneous temperature.

(a) 2 mm.

(b) 4 mm.

(c) 6 mm.

(d) 8 mm.

(e) 10 mm.

(f) 12 mm.

Figure 5.8: Sequence of tapered FHPP macrographs. Welds with increasing upsetting illustrate the

development of plasticised material and the change in shape of the frictional plane.

5-126

When the smoothening of the outer regions is almost complete, the plasticised

material produced in those regions starts to move towards the rotational centre. In

the outer regions almost no heat is produced during this intermediate phase,

because of the reduced frictional contact (the faying surfaces are now widely

smoothened in these regions). Lower radial forces and restricted material flow

conditions closer to the rotational centre combined with the constant generation of

plasticised material in the outer regions causes a material concentration on a circular

ring area around the rotational centre. Such a material concentration occurs after the

first 2 mm upsetting and can therefore be observed in the Figure 5.8(a). With

increasing saturation of plasticised material in the rotational centre, the ring area

starts to extend its diameter slowly towards the outer regions of the weld zone.

Temperature equilibrium is achieved, which is supported by a self-balancing effect.

These effects have been discussed by many authors [47,77,81,83-87] and can be

summarised as follows: As the temperature rises, the metal becomes more plastic

and the torque is reduced, leading to a lower heat generation with a resultant lower

temperature. This in turn softens the material, resulting in less deformation work and

in an instantaneous decrease of the temperature. The height of the plasticised zone

decreases and cooler areas now make frictional contact, heat up and become highly

plasticised. At this stage the consumable is fully plasticised across the bore of the

hole. The plasticised material develops at a rate faster than the axial feed rate of the

consumable rod, which means that the frictional rubbing surface (frictional interface)

rises along the consumable to form the dynamically recrystallised deposit material

with an almost constant upsetting rate, as demonstrated in Figure 5.8(b) to (e). The

plasticised material at the interface is maintained in a sufficiently viscous condition for

hydrostatic forces to be transmitted and therefore consolidate the joint at the

sidewalls. This process is maintained as long as the process control parameter (i.e.

upsetting) is not achieved, which stops the motor and allows the system to apply the

forging force to conclude the welding process (see Figure 5.8(f)). In Figure 5.8(c) and

(d) it can also be observed that although the frictional plane has not yet achieved the

middle depth of the hole, the consumable has deformed and already occupies the

cavity entirely at this point. During this stage, the stud material starts to be pressed

out of the friction area forming the flash. The white circular detail (see Figure 5.8(c))

and the arrows (Figure 5.8(d)) shows the top hole edge warped due to the further

upsetting imposed by the process after the filling of the hole.

5-127

Due to the intensive high temperature mechanical working undergone by the

consumable, the microstructure in the extruded material changes drastically. Figure

4.13 presents the grain size distribution after welding and clearly shows a reduction

in average grain size in those dynamically recrystallised areas. Generally, the main

advantage of a fine grain size is its enhanced mechanical properties, such as tensile

and yield strength. This relationship can be explained by the Hall-Petch equation (σy

= σo + kd-1/2, where σo is the yield stress of a single crystal, k is constant and d is the

grain size). The grain size effect on mechanical properties is more significant in alloys

that have an HCP structure, such as Mg and Ti, than in many other alloys with a

different crystalline structure (FCC or BCC), since there are fewer slip systems in the

HCP structure [167]. Other advantages of a fine grain size include the reduction in hot

tearing tendency and in the size of defects such as porosity, the improvement of

corrosion resistance and the machinability of cast products.

5.3 DISSIMILAR JOINTS – 8 MM DIAMETER/10 MM AND 20 MM UPSETTING

5.3.1 Process and Thermal History

Considering the results obtained from DAS combined with temperature data, some

conclusions can be drawn. Firstly, the upper part of the joint is generally 10 % to

15 % hotter than the lower, due to the higher frictional area and its corresponding

higher heat generation. Secondly, welds with higher upsetting and consequently with

longer welding times, presented the higher temperatures in both positions on

average. It suggests that longer welding times generate higher peak temperatures,

which could be associated with the heat input into the joint. Thirdly, one can observe

that in some cases joints with much longer welding times present lower temperatures

than joints performed within a shorter time. This statement contradicts the previous,

i.e. the heat input and its consequent peak temperature might not be directly derived

from welding time. An example of that, presented in Figures 5.9 and 5.10, is related

to samples AZAZ-a and d, which presents welding times of 51 s and 38 s

respectively, with correspondent peak temperatures of 467 °C and 469 °C, 390 °C

and 414 °C for upper (T1) and lower (T2) weld areas.

5-128

Figure 5.9: Peak temperature in upper thermocouple for similar and dissimilar welds, 10 mm and

20 mm upsetting.

In terms of material combinations, dissimilar welds always presented higher peak

temperatures when compared with those of similar welds. This fact may not be

directly related to the different thermal conductivity of the particular alloys, since the

AE42 and the MRI230D present higher thermal conductivity than that of AZ91 [168].

The fact that AE42 and MRI230D have a superior high temperature resistant

microstructure, and are consequently more stable, associated with the fragmentation

of the stable second phase precipitates in the case of MRI230D, which could

increase the heat generation rate, can be considered a reasonable explanation for

the higher temperatures observed during dissimilar welds. According to Figure 4.19,

MRIAZ and AEAZ have respectively achieved on average the highest temperatures

in the upper (472 °C) and lower (420 °C) weld areas for 10 mm upsetting joints. For

20 mm upsetting, the AEAZ joints reached the highest temperatures, with 502 °C and

471 °C respectively for the upper and lower weld areas. Thermocouple T1 shows a

variation of around 5 % between the highest (472 °C) and the lowest temperature

(448 °C), while T2 indicates a larger difference of around 11 % between both values

(420 °C for AEAZ and 374 °C for similar AZAZ welds) when 10 mm upsetting welds

are considered. For 20 mm upsetting, T1 shows values differing 4 % between the

highest (502 °C) and the lowest temperature (480 °C), while T2 points out a larger

difference of around 15 % between both values (471 °C for AEAZ and 402 °C for

similar AZAZ welds).

5-129

Samples of MRIAZ welded with 20 mm upsetting presented a particular feature in

relation to both the welding time and the peak temperature. According to Figure 5.10,

under the same welding conditions the preset upsetting is achieved in some cases

after 44 s (MRIAZ-d), whereas in another cases after 65 s (MRIAZ-b). On the other

hand, T2 indicates temperatures above 445 °C in some welds (MRIAZ-c) and around

390 °C in other cases (MRIAZ-a), which contribute to enlarge the scattering of

temperature measurement results. Based on these results, temperature variations up

to 55 °C within the same material combination were observed.

Figure 5.10: Peak temperature in lower thermocouple for similar and dissimilar welds, 10 mm and

20 mm upsetting.

As stated earlier, the difference in peak temperature indications may be correlated to

slightly different deposition mechanisms, but could also be attributed to the

positioning and type of the thermocouples as well as to the material itself. The first

problem that arises, when temperature measurement procedures are questioned, is

related to the accuracy of the depth of the thermocouple holes. It has been observed

that owing to the difficult procedure during the spark erosion of the thermocouple

holes, the actual distance between the tip of the holes and the bonding line varied

significantly. This variation, according to the literature, might vary the temperature

over a wider range [73,77,129,164,165], possibly causing an increased scattering in

temperature indications. Another possible source of uncertainty that has been

5-130

discussed recently is directly related to the type of thermocouple utilised.

Thermoelectric wires can undergo changes in their structure, because of sharp

bending, folding and stretching, which causes an error associated with the

inhomogeneities. In general, the hot junction – meaning the sensitive part of the

thermocouple – has to be located in the zone of the object where the temperature is

being measured. If this is not the case, the temperature is not measured correctly,

leading the readings to an error known as “errors because of wrong measuring

location” [169]. Further, errors associated with drift and with a wrong use of the

compensation cable are often reported and so should be also considered. According

to the manufacturer [169], acceptable deviations are around ± 2.5 °C up to 333 °C. In

this context, an error associated with the measurements should be considered and

the readings should not be taken as true absolute values. Furthermore, if joints

performed with the same parameters generate welds with time and temperature

values varying over a broad range, the deposition rate should be also investigated.

This variable, which can be monitored through the DAS, has shown differences

within the same group of welds. Such differences can be attributed either to oil flow

fluctuations, which would indicate a problem related to the welding system, or to

material imperfections. Figure 5.11 presents data recorded from welds MRIAZ-b and

d. As observed, both variables delivered by the welding system, i.e. rotational speed

and welding pressure, superimpose each other, indicating any interference of the

hardware on the joint formation is implausible. On the other hand, the deposition

(upsetting) rate, which is largely dependent on the material properties, seems to be

strongly affected by a microstructural imperfection related to the material. Figure 5.12

shows a micrograph of the MRI230D base material. As observed, even with lower

magnifications, dark regions representing grouped pores can be identified. As soon

as this region reaches the frictional interface during welding, the upsetting rate

(upsetting per time) increases instantaneously (see Figure 5.11/detail A). In addition,

the frictional torque reduces and the axial resistance decreases due to void spaces

within the material. Before the monitoring system adjusts the machine to keep to the

preset parameters, this occurrence causes a peak in the rotational speed (see Figure

5.11/detail B) and a momentaneous decrease in the welding pressure (see Figure

5.11/detail C). As demonstrated, a material imperfection may generate a series of

events, causing or contributing in some particular case to a great increase in the

measurement dispersion.

5-131

Figure 5.11: Weld plots of MRIAZ joints. Welds MRIAZ-b and MRIAZ-d present different upsetting

rates despite identically selected welding conditions.

Figure 5.12: Microstructure of the MRI230D base material. Black regions are voids, resulting from

chilled casting processes.

5-132

5.3.2 Metallurgical Characterisation / Joint Formation Mechanisms

5.3.2.1 AEAZ Joints

As described in the base material characterisation (see Section 4.1.2), the “as cast”

parent material is characterised by a pronounced dendritic structure in the matrix, in

combination with dispersed lamellar and particulate second phase particles of

Al11RE3, preferentially located along the grain boundaries [29,150,151]. Figure 5.13

demonstrates the microstructure across the interdendritic microstructure (in (a)) and

a complete rupture of the typical as-cast lamellar structure within the RZ in regions at

the vicinity of the BL on the AE42 side of the weld after welding (in (b)). Such a

rupture is to be expected, since the high level of strain produced in the weld

interface, generated by the relative movement between consumable and base plate,

associated to the welding pressure during the joining process, results in mechanical

fragmentation of the intermetallic Mn- and AlRE-based phases. However, in regions

towards the consumable centre, the microstructure seems to be to some extent

unaffected by the welding process. In those regions, the typical dendritic structure

with its lamellar second phase compounds remains almost unaltered, which suggests

that the consumable was not fully plasticised through its diameter. Micrographs of the

upper, unaffected BM in the middle of the RZ and of the second phase particles

fragmented in regions at the vicinity of the joint are shown in Figure 5.13 (see

indications “f” and “g” in Figure 4.20 for upper and lower stud areas respectively).

(a) (b)

Figure 5.13: Comparison of typical second phase particles across the interdendritic microstructure in

the upper unaffected BM (a) and in lower stud areas near the bottom bonding line (b).

5-133

According to the metallurgical characterisation (see Section 4.3.2.1), fragmented

particles have different sizes and are randomly distributed in the RZ along the joint

line. In particular, in regions along this line, mostly in the upper joint areas, clusters of

particles were observed as detailed by Figure 5.14. This particle concentration over

the bonding line appears probably due to the higher hardness of the intermetallic

compounds compared to the Mg matrix. The mechanical work and peak

temperatures are not enough to mechanically dissolve such particles and the hydro-

extraction effect, characteristic of friction processes, was not effective enough to

expel them from the welding interface. The particles are pressed together and

coalesce, due to the high temperatures/mechanical work imposed by the process.

Further, the thermal cycle undergone during the process is very short to promote

substantial diffusion and dissolution of the stable (Al-RE) intermetallic lamellar

compounds originally present in the BM.

Figure 5.14: Second phase particles concentrated along the bonding line of the dissimilar

AE42/AZ91D-T6 joint.

EDS analyses (see Figure 4.21 and Table 4.13) show lamellar particles within the

RZ, corresponding to points 1 and 2, having a high concentration of Al and RE with a

minor presence of Mg. Such compounds are suggested to be Al-RE based phases,

possibly Al11La3, Al4La or Al2Nd, all commonly present in Al-rare earth Mg alloys.

Although point 3 presents untypical Al content, points 3 and 4 reveal the average

5-134

composition of the AE42 α-Mg phase. Points 5 to 8 indicate the typical concentration

of the Mg-matrix in AZ91 alloys. Point 8, placed somewhere inside of the RZ, gives a

chemical composition very similar to that from the AZ91 matrix, indicating that a

transitional region with an increasing Al content was not formed. On the other side,

EDS analyses in points 9 and 10 recognise some dark grey regions with an abnormal

Al content in the base matrix. Since the typical as cast lamellae AE42 base material

structure was ruptured, associated with significantly higher temperatures, some Al-

based second phase particles possibly became locally molten and therefore partially

dissolved into the matrix. Due to the unstable solidification conditions imposed by the

rapid cooling rate, the free Al, previously dissolved in the Mg-matrix, might not have

had enough time to recombine and form stable intermetallic compounds. Hence

some higher Al-content regions were presumably formed.

Just as with the analysis done for the similar AZAZ welds, this section proposes the

mechanisms of joint formation for dissimilar welds, based on data from DAS,

temperature measurements and post-mortem microstructural analysis. As previously

mentioned, due to the differing material properties and structures within the ingots, to

some extent the formation of the joint may vary slightly, even under the same welding

conditions. Nevertheless, an attempt to describe the mechanisms involved in joint

formation is made in this section.

For the AEAZ welds, although Figure 4.18 demonstrated the welding time varying

over a relatively broad range, one can observe that the temperature remains at the

same level to some extent. Figure 5.15 shows the temperature cycles for identical

AEAZ welds with different upsetting rates, achieving, however, similar peak

temperatures. Vertical dashed lines indicate the end of the welding. Comparing

Figure 5.3, which shows the temperature progress for similar AZAZ welds, with

Figure 5.15 and keeping in mind the microstructural differences at the vicinity of the

joint and in the centre of the consumable, one can observe that the deposition

mechanisms and therefore the bonding formation are perhaps different in both cases.

Similar to that observed on the AZAZ welds, in the AEAZ joints the peak temperature

was always reached at the end of the weld. However, in these curves a smooth and

continuous temperature increase, without peaks or bends, was observed in both

thermocouple positions (T1 and T2).

5-135

Figure 5.15: Temperature monitoring in AE42/AZ91D-T6 welds performed with identical parameters.

Although AE alloys presented similar or even higher strengths to those of MRI or AZ,

AE studs have shown, unlike MRI and AZ, to be highly deformed during the welding

test under the same conditions. Considering the stress-strain curves presented in

Section 4.1, one can observe that the ductility of the AE alloys is clearly higher than

that of the MRI ones, which explains why the former material presents a higher

deformation under the same load. Bearing the high ductility of AE alloys in mind and

examining the temperature curves presented in Figure 5.15, it seems that the filling

mechanism observed in the similar welds, where the consumable member is fully

plasticised across the bore of the hole and through the thickness of the workpiece,

does not take place for this combination under these welding conditions. Instead, the

stud yields and occupies the cavity before a plasticised layer is created in this case

(see illustrations in Figure 5.16(a) and (b)), i.e. the consumable is inserted into the

cavity at a faster rate than the one at which the plasticised material develops. Since

the consumable is not plasticised, shear planes do not move up, generating a

constant temperature increase without peaks or breaks. After the hole has been

integrally occupied and the preset upsetting has not yet been achieved, the

exceeding material is pressed into the flash, as demonstrated in Figure 5.16(c). At

the final stages, the forging force is applied to consolidate the joint, which causes the

detachment of the flash material from the base plate (see Figure 5.16(d)).

5-136

(a) (b)

(c) (d)

Figure 5.16: Proposed bond formation mechanism of AE42/AZ91D-T6 welds.

The fact that the consumable is neither plasticised across the bore of the hole nor

through the height of the cavity perhaps causes the highest average peak

temperatures observed in Figure 4.19. Since the consumable is not plasticised, the

frictional area can be associated with the area of a conical frustum, which through its

larger surface area generates a higher amount of heat and consequently the highest

temperatures.

The bonding mechanism discussed above can be defined as a variation of friction

taper plug welding (FTPW), formerly developed as an alternative repair technique for

filling through the thickness of holes in offshore structures [110], however with a

bottom, i.e. as if the FTPW process would be performed inside a hole. In this

process, the heat generated from the contact between the consumable and the side

wall causes the material to soften and flow through the hole. At a subsequent stage,

the complete conical surface of the tapered plug is welded to the matching hole

almost instantaneously.

5-137

5.3.2.2 MRIAZ Joints

As demonstrated in Figure 4.22, for this configuration defect-free welds were also

produced, without visible discontinuities along the weld, for 10 mm and 20 mm

upsetting. The bonding line could be easily identified most of the time, not only

because of the different base material microstructures, but also due to the fine,

dynamically recrystallised microstructure in the MRI230D consumable, resulting from

the mechanical working imposed by the process. Figure 4.22(c) and (d) identified a

narrow region on the AZ91 side of the weld (TMAZ) where dynamically recrystallised

grains appear without the presence of the typical BM eutectic structure, which is also

dissolved in regions extending towards the HAZ (see overview Figure 4.22). The

absence of a eutectic structure in those regions can be explained as follows: The Al-

Mg phase diagram (see Figure 2.2) shows that a heating temperature higher than

390 °C causes the partial dissolution of the Mg17Al12 into the Mg-matrix of AZ91

alloys. Sufficient frictional heating introduced during the process causes the

dissolution of the eutectic phase in the transitional region (RZ). This free-eutectic

zone is extended to the HAZ in the upper joint regions, which suggests that these

regions are probably heated to temperatures higher than 390 °C during the process.

As also observed for the AEAZ joints, in regions close to the side walls, at the vicinity

of the joint line in the upper weld areas (see details in Figure 4.22(d)), detailed

metallographic investigations indicate the presence of grouped particles located

along the joint line. Such elongated particles are formed by the grouping of smaller

second phase particles, apparently fragmented during the welding process, which

could not be pushed out of the interface. Such Mg(Al,Ca) based particles maybe

have a superior strength than that of the matrix, in an intermediate stage of the

process being anchored and promoting subsequently a kind of stacking process,

which evolves and forms clusters of second phase particles in particular regions

along the joint line.

The “as cast” MRI230D ingot material is generally characterised by an α-Mg matrix,

some needlelike and polygonal type precipitates, described in the literature as being

Al8Mn5, Al2Ca and Al4Sr, dispersed within the matrix and by the Mg(Al,Ca) eutectics

with different morphologies located preferentially along the grain boundaries [33,40].

However, the microstructure on the MRI side of the weld is dominated by equiaxed

5-138

dynamically recrystallised grains, formed by deformation and the thermal cycle

imposed by the process. In particular regions, at both the vicinity of the joint and at

the middle of the consumable, a complete degradation of the typical base material

second phase particles Mg(Al,Ca) was observed. Under the same magnification,

Figure 5.17 highlights the common BM Mg(Al,Ca) intermetallic compounds, which

form an almost continuous network, distributed at grain boundaries and interdendritic

areas, and the resultant MRI microstructure after the welding. Fine Al, Ca, Sr and

other round and polygonal precipitates dispersed among the α-matrix seem not to be

strongly influenced by the process to any large extent, since they are present in the

deposit with approximately the same form and size as commonly found in the matrix.

Additionally, it should be mentioned that the morphology, i.e. size and form of the

final precipitates in the deposit, can be roughly controlled by the welding parameters.

(a) (b)

Figure 5.17: Microstructural changes in the consumable in dissimilar MRI230D/AZ91D-T6 welds. In

(a) the typical Mg(Al,Ca) eutectics along the grain boundaries and in (b) the same material after

welding at mid-thickness at the rotational centre (see position “G” Figure 4.22).

As observed during the metallographic analysis, the extruded region presents an

inhomogeneous microstructure along the deposit height in both welding conditions

(10 mm and 20 mm upsetting). In comparison with the MRI base material, where

grain size analyses indicate values ranging from 80 μm to 120 μm, the deposit’s grain

size generally varies from a finer structure at the bottom, with an average grain

diameter ranging from 5 μm to 13 μm, to a coarser structure at the top, where values

varying from 18 μm to 22 μm were identified. Generally, the welds performed with

20 mm upsetting presented a more expressive grain growth in the upper regions of

5-139

the extruded area, probably as a consequence of the slightly higher temperatures

and also the longer welding times, which allows not only recrystallisation, but also

grain growth processes to take place subsequently. Further, Figure 5.18 evidences

grain size differences in regions diametrically opposed (see Figure 4.22 indications

“f” and “h”), theoretically where the welding conditions imposed were similar and

therefore similar final structures should be obtained.

(a) (b)

Figure 5.18: Grain size assessment within the extruded material. In (a) the microstructure in the “f”

(average diameter: 7.3 μm) and (b) in the “h” (average diameter: 11.8 μm) position.

North et al. [164] reported that transient local melting occurs sufficiently often during

Al7075-T6 spot welding and directly affects the stir zone formation. Bearing this in

mind, the high strain rates imposed by the process, associated with local liquid phase

formation, in particular regions provide the conditions for grain boundary sliding,

which influence the RZ’s formation, by reducing or increasing the local heat

generation. The above mentioned mechanism varies from region to region,

dependent on the base material’s homogeneity and maybe dictates and generates

the resultant microstructure inside the extruded area.

EDS analyses (see Figure 4.23 and Table 4.14) along the welding line indicate the

presence of different intermetallic compounds. Round particles within the AZ91 side

of the weld, corresponding to point 1, showed a high concentration of Al and Mn with

a minor presence of Mg. Such compounds are suggested to be AlMn based phases,

commonly present in AZ alloys and described in the literature as being the Al8Mn5

phase. Point 2 indicates a high concentration of Mg, Al and Mn with some evidence

5-140

of Ca, Zn and Sr. In AZ, as well as in the MRI base material, some binary, e.g. AlCa

(MRI) and AlMn (AZ and MRI), ternary, e.g. MgAlCa and even quaternary, e.g.

MgCaAlSb, MgCaZnSn and MgCaCuSn (MRI) phases are often identified. However,

phases with Mg, Al and Mn with some other minor presences seem to be created by

the conditions generated in the welding process. A further and precise investigation,

also involving an XRD analysis, would be helpful in determining and identifying such

phases. Points 3 to 6 basically reveal Mg(Al,Ca) second phase particles, possibly

originated from the mechanical fragmentation of the eutectics located previously

along the grain boundaries of MRI BM, however, with main element contents varying

over a broad range. Scans corresponding to points 7 and 8 show the composition of

the MRI matrix in regions near to the bonding line, whereas points 9 to 11 identifies

further fragmented precipitates on the MRI side, suggesting that such phases vary,

not only in terms of content, but also in relation to the presence or absence of a

specific element.

Conversely, joints performed with MRI studs seem not to follow the same joint

formation mechanisms determined for AEAZ welds. According to the evidence

presented in the metallographic analysis (see Figures 4.22, 5.17 and 5.18) and

based on the temperature profiles recorded during 10 mm (MRIAZ-2) and 20 mm

(MRIAZ-d) upsettings of MRIAZ welds (see Figure 5.19), one can affirm that, within

this combination, the joint formation follows the same sequence discussed in Section

5.2.2 for similar AZAZ welds. Due to the matching correlation material/welding

conditions, the plasticised layer is generated through the consumable’s diameter, and

rises along the hole, leaving beneath a fine dynamically recrystallised deposited

material. Similar to that observed for the AZAZ joints, at T2 the temperature

increases, experiencing a short drop as the frictional interface achieves the level of

the thermocouple hole and slightly rises again towards the end of the process. At T1

the temperature increases continuously, achieving the maximum value and

remaining practically constant up to the end of the process. In this case the self-

regulating peak temperature mechanism proposed by North et al. [166] and detailed in

Section 5.2.1, probably took place and forced the temperature not to overcome the

solidus line. A temperature drop was not observed at T1, since the peak temperature

is achieved somewhere before the frictional interface reaches the upper

thermocouple level.

5-141

Figure 5.19: Temperature monitoring in MRI230D/AZ91D-T6 welds.

5.3.2.3 AZAZ Joints

As also observed with the other combinations, after some preliminary trials sound

welds were obtained with no bonding defects along the welding line and without

porosity. In conventional fusion welding of AZ cast alloys, porosity is always an

important issue to be considered, since the gases entrapped during the casting

process are liberated and remain in the welding seam under certain conditions [170].

Micrographs from the bottom of the weld (see Figure 4.24(b), (c) and (d)) show a

broad range on both sides, where a dynamically recrystallised structure is observed,

which enables the severe plastic deformation in the weld zone to be accommodated

by the flow of the material in the solid state. Hence, DRX facilitates the solid-state

flow and enables a suitable joint to be produced by FHPP. Also to be considered is

the heat arising from the deformation-induced DRX that contributes to the higher

temperatures achieved in the weld zone, and consequently to the formation and

consolidation of the joint. Further, the intercalated microstructures at the bottom,

seen in Figure 4.24(b) and (d), with its lamellar-like shear bands have been

previously reported in other studies involving friction stir welding of Mg alloys [171]. In

such regions one cannot observe a sharp demarcation, where it passes from base

plate to rod material. Instead, there is a gradual transitioning into the weld zone

dictated by DRX mechanisms implemented throughout the friction process.

5-142

The microstructure of AZAZ samples in regions around the final frictional plane in the

upper welding area (see Figure 5.20) is characterised by complex intercalated flow

patterns. In these areas, part of the former base material is recrystallised, while α-

matrix islands, without the characteristic lamellar Mg17Al12 eutectics, can be observed

in between. In such regions, it seems that the energy supplied by the process was

not sufficient to cause a complete plasticisation with subsequent recrystallisation.

Instead, the heat input provides enough energy to dissolve the β-phase and to

nucleate new grains, generating a network, however in islands within the matrix.

Figure 5.20: Microstructure around the final frictional plane (see point “F” in Figure 4.24).

Welds with higher upsetting generally showed a larger grain size within the extruded

area. Higher upsetting and the consequent longer welding/cooling times certainly

play the most important role in determining the final weld properties. However, in

terms of joint quality, both conditions visually seem to produce welds with adequate

joint strength.

EDS was performed, with the investigated points indicated in Figure 4.26 and the

corresponding results shown in Table 4.15. Figure 4.26(a) shows round particles

stacking at the sidewall interface, corresponding to points 1 and 2. EDS analysis of

these particles indicates a high concentration of Al and Mn with a minor presence of

Mg and Si. Such compounds are suggested to be AlMn based phases, commonly

5-143

present in AZ alloys and described in the literature as being Al8Mn5. The

concentration of such particles over the bonding line should be controlled by welding

conditions, in order to fragment such phases and so avoiding areas with low ductility

in the final component. Points 3 to 5 (see Figure 4.26(b)) show a high concentration

of Mg and Al with some evidence of Zn in scans over the grain boundaries. Such

regions, probably formed by Mg17Al12 phase, are a strong indication of local melting,

which causes low melting-point intermetallic eutectics, concentrated on the grain

boundaries, to be segregated, molten and re-precipitated during the cooling phase.

Temperature profiles and investigations involving the mechanisms of bonding

formation led to results very similar to those presented at Section 5.2.2. Therefore

such mechanisms will not be further discussed.

5.4 MECHANICAL CHARACTERISATION

5.4.1 Hardness Tests

As shown in Figure 4.15, AZAZ-06/12 (series of welds performed with 6 mm diameter

studs and 12 mm upsetting) welded joints have resulted in generally similar hardness

profiles along the transversal cross-sections, with no softening or hardening effects

independent of the welding parameters. Although dwell times and cooling rates are

seen to be different, peak temperatures experienced by the welds were to some

extent similar. Therefore, and based on the metallurgical analysis, the resultant

microstructures within the extruded area were similar, which would consequently

generate similar properties within the RZ. Compared to the BM, the possible increase

in hardness as a consequence of the fine dynamically recrystallised microstructure,

resulting from the plastic deformation imposed by the process, is to some extent

compensated by the dissolution of the second phase particles, which are produced to

increase the strength of such alloys. Both phenomena influence one another and

probably culminate in forming a weld deposit with an analogue of the hardness level

of base material. Johnson et al. [5] and Kato et al. [114] have reported similar results,

working respectively with friction stir welding and friction welding of Mg-Zn-Al-Mn

alloys. According to their work, there did not appear to be any significant hardening,

which can also be seen in the scans performed across the welded regions.

5-144

For AZAZ-06/12 and for AZAZ-08/10-20 (series of welds performed with 8 mm

diameter studs and 10/20 mm upsetting), scans performed in the lower weld areas

presented a tendency to produce higher hardness values than those from the upper

weld areas (see Figures 4.15 and 4.27(e) and (f)). The slightly higher hardness in

those regions can be attributed to the smaller grains (see Figure 4.13 and Section

4.3.2.3), generated due to the faster cooling rates imposed at the initial stages of the

process. Draugelates and Schram [118] reported higher hardness values along the

bonding line in rotational friction welding of AZ alloys. According to their work, the

significant grain refinement imposed by the process suppresses the effect caused by

the Mg17Al12 phase dissolution, thereby generating hardness values approximately

20 % higher in the welding area. Working with an Mg-Al-Zn friction welded alloy,

Ogawa et al. [113] also reported an increase in hardness along the welding line, as a

consequence of the work-hardening effect sustained under a fairly high forging

pressure. Friction stir welds using the same alloys (AZ31) have indicated either

higher hardness values across the welding zone, due to the refining mechanism

imposed by the process [172], or lower hardness, due to the softening effect caused by

the thermal cycle during the process [173]. According to the latter work, since AZ31

derives much of its strength from being rolled as a wrought product, the temperature

adversely affects its properties, thereby causing such a softening effect [173].

In the AEAZ joints, material hardening in comparison to the BM is clearly observed

from start to finish within the deposit (see Figure 4.27(a) and (b)), resulting from the

welding thermal cycle imposed by the process. Although in this particular case the

stud material was not fully plasticised across the bore of the hole, grain refinement,

as a consequence of the mechanical working, has occurred and seems to elevate the

average hardness. Pinheiro et al. [73] observed for similar AE42 rotational friction

welds, a bonding line characterised by a thin layer of dynamically recrystallised

grains with slightly higher hardness values. However, hardness values in the RZ

were practically unaltered by an upsetting increase, suggesting that thermal cycles,

although slightly different, affect the microstructure in a similar way. Bowles et al. [112]

reported without further discussion the feasibility of friction welding to produce good

quality dissimilar AEAZ joints with no significant loss in hardness across the welding

area.

5-145

Similar to the previous case, the welding process and the consequent production of a

very refined microstructure seems to influence the hardness values in the RZ of the

MRIAZ welds. Scans performed near the top surface (see Figure 4.27(c) and (d)) are

very smooth and present a lower scattering, probably due to the higher heat

generated in those locations. Since the formation of these areas and the resultant

microstructure in 10 mm and 20 mm-upsetting welds are to some extent similar, a

hardening mechanism, probably involving fine precipitate dissolution or re-

precipitation during cooling, took place under these conditions.

Generally, the scans performed at 2 mm from the top surface presented a much

smaller scattering when compared with those carried out across the lower stud areas.

This behaviour can be attributed to the partial dissolution/fragmentation of the

eutectic phases within the extruded material and to the dissolution of the lamellar β-

phase in the HAZ on the AZ91 side of the weld, due to the high temperatures

achieved in those locations. Within the extruded area (RZ) the scatter is reduced,

obviously due to a much more homogenised microstructure, generated by the hot

working imposed by the welding process. Since such high hardness products are

dissolved, the scatter is reduced. However, in scans performed across the bottom

weld area, the decrease in dispersion is easily observed as soon as the extruded

area is reached. In fact, as observed in the metallurgical characterisation (see

overviews in Figures 4.20, 4.22 and 4.24), in the bottom areas on the AZ91 side of

the weld, the BM is not modified in regions very close to the BL into a deep

extension. Since the workpieces are at room temperature at the beginning of the

process, the generated heat is almost instantaneously conducted out of the welding

region. Hence the achieved temperatures, as well as the dwell times in these

locations, are not high enough to partially homogenise the microstructure, causing

the profiles to show a significantly increased scattering.

5.4.2 Pull-out Tests

According to the results shown in Sections 4.2.4.2 and 4.3.3.2, pull-out tests in the

joints that were produced using a broad range of parameters have confirmed the

ability of the process to deliver high-strength welds. After 10 kN is achieved, which is

equivalent to a stress of about 32 MPa at the cross section of the stud, yielding

begins. From that point, the pull-out test can be compared with a standard tensile

5-146

test, in which the sample has half of the gauge length. Based on that, the maximum

achieved load seems to match the BM properties and small variations can be

considered acceptable, due to the broad range in which BM properties vary.

Converting pull-out values to strength, AE, MRI and AZ studs were loaded up to

170 MPa, 113 MPa and 198 MPa respectively, which to some extent matches the

values found for the BMs (see Section 4.1).

Bearing in mind the application, in which the process is planned to be used, pull-out

tests confirm the suitability of the proposed concept for locally reinforcing Mg

components with high quality welds. According to the VDI-Handbuch Konstruktion [174], shank bolts are maximally loaded at their minimum yield point. Based on that

and on the results delivered by pull-out tests (see Sections 4.2.4.2 and 4.3.3.2), for

AZAZ-06/12 welds, steel bolts with a minimum of 10 mm, 12 mm and 14 mm nominal

diameters respectively from 12.9, 10.9 and 8.8 strength grades could be used at

room temperature without collapsing the joint, i.e. without pulling the welded stud out

of the base plate. If 08/10-20 welds are considered, the results delivered by pull-out

tests indicate that steel bolts with a minimum of 10 mm, 12 mm and 14 mm nominal

diameters from 8.8 strength grade could be used with MRIAZ, AEAZ and AZAZ joints

respectively at room temperatures without causing the joint to fail. Since an M8 screw

strength of grade 8.8 is the largest bolt actually in use in the automotive industry [9,10],

within the spectrum of applications delineated for the present study, it can be

affirmed that these joints would be able to carry the load with some margin of safety.

5.4.3 Transversal Tensile Tests

Tensile tests from AZAZ-06/12 welds demonstrated joints with good performance,

particularly those performed with higher pressures. Figure 4.17 shows that joints

performed under different welding conditions have generally similar mechanical

properties to those of the base material. Although joints welded with the highest

pressure presented a very inhomogeneous microstructure in the metallographic

analysis, in the transversal tensile test these samples presented the highest average

strength. Low pressure joints achieved strength values in the range of 87 % to 98 %

of the BM tensile properties for lower and upper weld regions respectively. However,

high pressure welds presented values even superior to those of the BM values. It

seems that in the former, the magnitude of both the welding pressure and

5-147

subsequently the forging force was not high enough to bring stud and base material

into the intimate contact necessary to produce a strong bond. In his experiments,

Meyer [78] demonstrated a decreasing influence of the welding pressure and of the

forging force on the joint formation at the bottom of the cavity during the last stages

of the welding cycle. Especially in welds performed at lower pressures, during the

final stages of the welding process the microstructure, and consequently the bond

formation at the bottom area, are to some extent already consolidated and therefore

do not suffer a significant influence over the events occurring in the upper region of

the joint. This fact could explain, as demonstrated in Table 5.2, why most of the

samples welded with lower pressures collapsed at the bonding line, particularly in

trials carried out on the lower part of the joint. However, with increasing welding

pressures, the nominal set value is totally transferred to the lower joint parts, as long

as the limiting control parameter is not reached. Table 5.2 mirrors this tendency,

since at higher pressures (samples 2 and 3) most of the joints failed in the base

material, outside the extruded area.

Table 5.2: Position and frequency of failure after tensile testing of AZAZ-06/12 joints.

Failures at the BL Failures in the BM

Upper 37.5% 62.5% MgAZAZ - 1

Lower 75% 25%

Upper 12.5% 87.5% MgAZAZ – 2

Lower 12.5% 87.5%

Upper 0% 100% MgAZAZ – 3

Lower 0% 100%

Tensile properties of dissimilar welds also indicated that high performance joints can

be obtained using FHPP. The average strength of samples from upper and lower

stud areas is shown in Figures 5.21 and 5.22 respectively. Based on these results,

and on the curves presented in Figure 4.29, one can observe that the tensile

behaviour tends to match the AZ91 BM curve, since this material composes the

largest extension of the tensile specimen. Furthermore, grain refinement associated

with particle fragmentation and a porosity-free microstructure, due to the hot working

imposed by the process, seems to contribute to the enhancement of tensile

properties. Table 5.3 indicates joints welded with AE/MRI inserts having up to 90 %

superior strength of the BM in some cases. The failure location varied randomly, but

in most cases occurred in the BM, outside the extruded area.

5-148

Figure 5.21: Effects of upsetting on the yield strength, tensile strength and elongation at rupture in

8 mm/10-20 mm welds (upper samples) in comparison with results from the base material.

Figure 5.22: Effects of upsetting on the yield strength, tensile strength and elongation at rupture in

8 mm/10-20 mm welds (lower samples) in comparison with results from the base material.

According to Figures 5.21, 5.22 and Table 5.3, AEAZ welds presented generally the

highest elongation, probably due to the AE42 base material’s properties. However, in

some cases, as demonstrated for welds AEAZ-10 mm, in the upper weld areas the

concentration of second phase particles along the BL (see Figure 5.14) seems to

decrease the elongation abruptly and often causes the fracture to take place in these

regions. As reported on the literature [175] and observed in some initial SEM analysis,

5-149

such clusters of particles along the bonding line may have increased the notch

sensitivity in these particular areas, leading the joint to collapse with lower

elongation, but with significantly higher strength values. Tensile strength values

varied arbitrarily, without a defined relation to both the specimen location (upper and

lower samples) and the upsetting.

MRIAZ welds also presented satisfactory strengths and the resultant refined

microstructure seems to contribute in producing such a high performance. In this

case, the forging force seems to play an important role, since upper weld area

specimens presented clearly higher strengths when compared with lower specimens.

As discussed by Meyer [78], during the final stages of the FHPP welding process the

microstructure of the bottom area is to some extent already consolidated and

therefore suffers a marginal influence of the forging force. Furthermore, MRIAZ welds

with 20 mm upsetting presented a more expressive grain growth in the upper weld

regions (see Section 5.3.2.2), which probably contributed to generating the observed

inferior tensile strengths in these locations. In the MRIAZ-10 mm upsetting welds, the

failure was mainly located in the BM, in a region outside the RZ. On the other hand,

the failure location moved towards the RZ with higher upsetting welds, which can be

attributed to some extent to the different BM properties. Table 5.3 lists the highest

absolute values and shows the feasibility of the process to perform high-strength

dissimilar MRIAZ welds.

Transversal tensile samples from AZAZ-08/10-20 welds seem generally to present a

higher elongation in relation to base material, probably due to the formation of a

refined DRX microstructure in the extruded area. In most of the cases the failure

location occurred in the BM outside the RZ. Generally, 20 mm upsetting specimens

seem to present slightly higher strengths, although in the lower stud areas there is a

global tendency (observed in all the combinations) to equalise the tensile strength for

10 mm and 20 mm upsetting values. The elongation, as observed also for AEAZ and

for MRIAZ combinations, was significantly higher in samples from the upper weld

areas. This characteristic can be attributed to the larger specimen gauge length

occupied by the refined microstructure produced by the welding process. Moreover,

AZAZ joints tested within this group demonstrated a similar behaviour to the AZAZ-

06/12 welds (see Section 4.2.4.3).

5-150

Table 5.3: Maximum tensile strength experienced by 08/10-20 mm joints.


Upper 184 132 1.32 AEAZ-10 mm

Lower 183 132 1.32

Upper 175 100 3.83 AEAZ-20 mm

Lower 155 111 1.73

Upper 186 127 1.76 MRIAZ-10 mm

Lower 155 128 1.01

Upper 167 106 2.48 MRIAZ-20 mm

Lower 160 129 0.96

Upper 182 111 2.70 AZAZ-10 mm

Lower 175 161 0.61

Upper 188 112 2.29 AZAZ-20 mm

Lower 168 104 1.86

AE42 Base Material 104 ± 8 90 ± 2 4.5 ± 1

MRI230D Base Material 94 ± 8 90.1 ± 2 0.8 ± 0.2


5.5 JOINT PERFORMANCE

5.5.1 Creep Tests

Following the procedure presented in Section 3.5.7, the creep resistance of the

materials used within this study was determined (see Section 4.4.1). Although MRI

alloys have confirmed their superior performance, some results were to some extent

unexpected and widely differed from values reported in the literature. As

demonstrated in Figure 5.23, Mg-Al-RE (AE) alloys presented at 125 °C and 150 °C

a significantly lower creep resistance when compared with Mg-Al-Zn (AZ) alloys.

While AE alloys show strain values of 1.8 % and 4.1 % for 125 °C and 150 °C

respectively, the latter indicates values of 0.6 % and 2.0 % under the same testing

conditions, i.e. at 80 MPa and after 100 testing hours. As presented in Section 3.1,

Mg AZ91D was heat treated and artificially aged to the T6 condition. The standard T6

heat treatment causes partial homogenisation of the microstructure and at the same

time prevents, at least to some extent, precipitation of the intergranular β-phase

particles. After solid solution treatment, most of the β-phase is dissolved in the

matrix, but during aging treatment a trace amount of β-phase precipitates, again

along the grain boundaries (see Figure 4.1(c)). Due to the subsequent aging, AZ

5-151

alloys presented at 125 °C a higher stability with a consequently better performance

by comparison with AE42 alloys. At 150 °C, however, the coarsening of the Mg17Al12

phase precipitated along the grain boundaries seems to decrease the creep

resistance of AZ alloys. Therefore, at this temperature, the differences between the

performance of AZ and AE alloys are reduced. Bearing in mind that magnesium

alloys undergo creep mainly by grain-boundary sliding, at 175 °C the coarsened

Mg17Al12 phase, which has a melting point of approximately 460 °C and is

comparatively soft at lower temperatures, does not serve to pin the grain boundaries

in AZ alloys. As a consequence, Mg-Al-Zn alloys creeps very early under those

testing conditions.

In AE alloys, it is estimated that the presence of one or more stable Al-RE and Mg-

RE phases contributes greatly to its superior behaviour at elevated temperatures,

through intergranular and transgranular strengthening. However, as demonstrated in

Figure 4.3(a), the intermetallic Al-RE phase in the ingot material is inhomogeneously

distributed within the matrix. Furthermore, the matrix is composed of bulky

precipitates with a large relative distance to each other, which may contribute to the

lower creep resistance observed in AE alloys. Working with a die-cast Mg-Al-RE (AE)

alloy, Bakke et al. [140] reported a tensile creep strain of 0.1 % after 100 hours at

175 °C/75 MPa, whereas Dieringa [27], working with squeeze casting materials found

compressive creep strain values of around 3 % under similar conditions. In both

cases, as may be expected, their reported BM creep strain values are clearly lower

than those determined within this study.

MRI alloys demonstrated the highest performance among the BMs. As reported in

the literature [43], the formation of eutectic Mg17Al12 is suppressed by the presence of

Ca and Zn. The resultant Mg-Al-Zn-Ca phases are evenly distributed along the grain

boundaries and impede grain boundary sliding. Thus, superior behaviours of AXJ

and MRI alloys were observed.

5-152

(a)

(b)

(c)

Figure 5.23: BM compression creep curves at testing temperatures.

5-153

For AEAZ joints, creep resistance of the welded area indicated at 125 °C and 150 °C

similar strain values to those determined for the upper (ref.) area (see Figure 4.30).

However, at higher temperatures, where testing conditions become very severe for

Mg alloys, the upper (ref.) specimen showed a clearly superior creep resistance.

Based on the mechanism of joint formation identified for AEAZ welds, one could

observe that minor differences occur between microstructures of deposited and base

material, because the stud is not fully plasticised across the bore of the hole.

According to Moreno [44], a relatively refined microstructure, resulting from the hot

working imposed by the process, would be detrimental to high temperature

properties, since the creep of metals is driven by grain boundary sliding mechanisms.

However, since no expressive grain refinement was observed for AEAZ joints,

mechanical fragmentation of the second phase particles in regions close to the

bonding line seems to be a more reasonable explanation for the lower creep

resistance of the welded areas. Figure 5.24 demonstrates lower (A) and upper (B)

samples after testing at 150 °C/80 MPa. As evidenced in the picture, the deformation

of the welded sample (A) is concentrated in a small area, while in the reference

sample it is distributed along the whole specimen length. Knowing that the deformed

area coincides with the bottom weld area, it seems that mechanical fragmentation

somehow influences the creep mechanisms, decreasing the resistance of the

material locally. Figure 5.13 shows a comparison of typical second phase particles in

position 1 (upper weld areas – see Figure 5.13(a)) and position 2 (lower stud areas

near the bottom bonding line – see Figure 5.13(b)) of sample (A) in Figure 5.24.

Figure 5.24: AE lower (left - A) and reference (right - B) samples after creep test at 150 °C/80 MPa.

5-154

Bearing in mind that AE42 alloys have their high temperature strength attributed to

the formation of stable AlRE precipitates, which suppress the formation of β-phase, it

could be that the mechanical fragmentation caused by the process liberates some Al

within the matrix. This free Al would hence increase the probability of Mg17Al12 phase

formation, which would consequently contribute to a decrease in high temperature

properties. EDS analyses performed during metallurgical characterisation of AEAZ

joints identified some dark-grey regions with an abnormal Al content (see Figure 4.21

and Table 4.13, points 9 and 10).

In addition, the calculated activation energy of 121.9 kJ/mol and 78.2 kJ/mol

respectively, for lower and upper (ref.) samples (see Figure 5.25), are to some extent

in agreement with results published by Dieringa [27] for AE42 and AE42+saffil

reinforced alloys. The lower activation energy determined for upper reference

samples may be related to the limited number of experiments performed.

Furthermore, the activation energy estimated from the plot is lower than that for

lattice self-diffusion in Mg (135 kJ/mol) or activation energy for diffusion of Al in Mg

(143 kJ/mol). Since these values give an estimation of the predominant creep modus,

one can conjecture that both samples present different deformation mechanisms. In

order to calculate the stress exponent “n” and determine accurately the deformation

mechanisms acting, tests under different stresses should be performed, which are

outside the scope of the present work.

Figure 5.25: Plots of ln(d/dt) against 1/T for AE42 reinforcements in order to determine the activation

energy for creep. Test conditions: 125 °C, 150 °C and 175 °C, under 80 MPa.

5-155

As indicated in Section 4.4.1.2, MRI alloys presented a very good creep

performance. Despite this, the creep values ascertained within this study are far

removed from those reported in the literature [176,177]. At 175 °C, for example, the

upper reference sample creeps at around 3 %, while the lower sample creeps

between 9 % and 10 % after 100 hours at 80 MPa. In the literature, creep strains

varying between 0.02 % and 0.15 % for Mg-Al-Sr-Ca alloys [176] and around 0.25 %

for MRI230D alloys [177] have been reported after 100 and 200 hours respectively at

175 °C/70 MPa. Considering that the MRI230D is a fairly newly developed alloy,

there are a limited number of publications involving its high temperature properties

and the microstructural changes related to it. This lack of information in the literature

makes a relevant analysis of the mechanisms acting during creep tests of Mg-Al-Ca-

Sr alloys a relatively difficult task.

According to Hirai [178], additions of Ca and Sr, up to 1wt% and 0.5wt% respectively,

in AZ alloys decrease the grain size substantially, consequently improving the creep

properties. Bearing that in mind and considering the expressive grain size refinement

within the extruded material, a higher creep resistance would be expected in those

areas when compared with the reference samples. However, similar to that observed

in regions surrounding the bonding line in AEAZ welds, the process imposes a

complete disrupture of the second phase precipitates (see Figure 5.17, Section

5.3.2.2), which apparently liberates Al into the α-matrix (compare Table 4.8, EDS

points 7,8 and 9 with Table 4.14, EDS points 7 and 8). Although the free Al inside the

matrix is reported to form stable Al4Sr and Al2Ca [33], its excess probably aids the β-

phase formation, but which perhaps justifies the reduction of creep properties inside

the extruded material, as demonstrated in Figure 4.31. This effect seems to exceed

the positive contribution of grain refinement. Another possible reason for the reduced

resistance within the deposit was reported by Bai et al [176]. According to their study,

the improvement of creep resistance caused by Sr and Ca additions may be

attributed to the formation of an intermetallic network at grain boundaries.

Considering grain boundary sliding the most common creep mechanism in alloys

based on Mg-Al containing Ca and Sr, the intermetallic network constitutes a

restriction, thereby improving the overall creep resistance over different temperature

and stress conditions. According to this theory, the network fragmentation imposed

by the process would have a detrimental influence on creep properties within the

5-156

deposit, which is to some extent confirmed by Figure 4.31. Additionally, it has also

been reported [176] that intermetallic particles present in Mg-Al-Sr-Ca quaternary

systems are larger than 1 μm and are therefore not effective on inhibiting dislocation

motions. Metallurgical characterisation (see Section 4.3.2.2) shows that fine

polygonal Al, Ca and Sr precipitates remain apparently unchanged after the weld

operation and consequently were not related to changes in the creep properties.

Further TEM microscopic analysis would be helpful, perhaps to comprehend the

changes on fine precipitates and their probable relationships with creep properties

after the welding operation.

Moreover, the activation energy for creep was calculated and reaches 110.1 kJ/mol

and 109.9 kJ/mol respectively for lower and upper (ref.) samples (see Figure 5.26).

Compared with the activation energy found in the literature [27,179], the observed

values seem to be of the same order of magnitude, indicating that on average the

same energy is necessary to initiate the creep process in both, the upper and lower

samples.

Figure 5.26: Plots of ln(d/dt) against 1/T for MRI230D reinforcements, in order to determine the

activation energy for creep. Test conditions: 125 °C, 150 °C and 175 °C, under 80 MPa.

5-157

Since AZ91 is one of the most used Mg alloys, many authors have been investigating

its high temperature properties [179-181] recently. According to Figure 4.32, samples

extracted from the lower stud areas of similar AZAZ joints presented, in all three

cases, a reduced creep resistance when compared with the upper (ref.) samples. As

demonstrated in Sections 4.2.3 and 4.3.2.3, in similar AZAZ welds, the extruded

material shows a very refined dynamically recrystallised microstructure with a

complete rupture of the intragranular lamellae β-phase and a further re-precipitation

of these phases at grain boundaries, which can explain its poor creep properties. On

the one hand, grain refinement associated with the presence of intragranular

lamellae β-phase has been reported to result in improved creep resistance for alloys

that contain a hard divorced eutectic, since the fully grain-refined state with a hard

divorced eutectic helps to lock grain boundaries and reduce grain boundary sliding [182]. On the other hand, discontinuous precipitation of Mg17Al12 at grain boundaries

allow grain boundary sliding at elevated temperatures, increasing creep susceptibility [180]. In creep tests of AZAZ welds, the effect of grain refinement seems to be

suppressed by the rupture of the intragranular lamellae β-phase associated with its

precipitation along grain boundaries. As a consequence the welded region presents a

reduced creep resistance.

As observed within this section, the authors have different opinions concerning the

influence of grain size and the presence of second phase particles in the creep

properties of Mg alloys. At temperatures above half the melting point, deformation

can occur in different ways. A rough way of distinguishing when grain-boundary

sliding becomes prominent is by the equicohesive temperature. Above this

temperature the grain boundary region is weaker than the interior and strength

increases with an increase in grain size. Below the equicohesive temperature the

grain-boundary region is stronger than the grain interior and strength increases with a

decrease in grain size (thereby increasing the grain boundary area). In this context, a

determination of the creep mechanism would be fundamental to define the welding

conditions aimed at the enhancement of joint performance. Furthermore, the

activation energy was calculated from Arrhenius plots reaching 123.5 kJ/mol and

147.7 kJ/mol for lower and upper (ref.) samples respectively (see Figure 5.27). This

contrasts with the activation energy values ranging from 121 kJ/mol to 171 kJ/mol as

found by Guo et al. [179], with activation energy for lattice self-diffusion in Mg

5-158

(135 kJ/mol) or activation energy for diffusion of Al in Mg (143 kJ/mol). Thus the

observed values are somewhat in a reasonable range. In the present work, the

calculated activation energy for creep exhibits a clear increase in upper samples,

which totally agrees with the higher creep resistance tendency of upper samples

shown by creep curves.

Figure 5.27: Plots of ln(d/dt) against 1/T for AZ91D-T6 reinforcements in order to determine the

activation energy for creep. Test conditions: 125 °C, 150 °C and 175 °C, under 80 MPa.

In addition, a brief comparison between the creep resistance of the three welded

combinations and the AZ91D-T6 base material under different testing conditions was

carried out (see Figure 5.28). Considering the particularities of the materials

specifically used within this study (variable tensile strength, presence of voids, etc.)

at lower temperatures one can observe a similar behaviour between AZ91 and

MRI230D. However, at 175 °C the superior high-temperature properties of MRI230D

become evident. Interesting in this context was the behaviour of the AE inserts,

which always presented after welding an inferior creep resistance when compared

with AZ91D-T6, even at higher temperatures. According to experiments carried out

by Baake et al. [140] and by Aune and Ruden [36] creep strength of AE42 at 100 °C,

150 °C, 175 °C and 200 °C under different stresses always presented a superior

resistance in comparison with AZ91. Changes imposed by the welding process

cannot be used to justify the poor creep behaviour of AE42, since such differences

were also evidenced in the reference samples (see Figure 5.23).

5-159

(a)

(b)

(c)

Figure 5.28: In (a), (b) and (c), the compression creep curves are plotted for similar and dissimilar

welded samples at 125 °C, 150 °C and 175 °C respectively. Curves AZAZ REF were also plotted at

different temperatures and represent the AZ-BM behaviour for further comparison.

5-160

Considering the objectives outlined within this study and the results of the creep

tests, Mg-Al-Ca-Sr (MRI) alloys seems to be the most suitable to locally reinforce

AZ91 components. Although high strength welds were produced, AE42 seems not to

significantly contribute to increase the high temperature properties of AZ91 alloys

locally. Furthermore, AZ91D-T6 inserts, which could improve high temperature

properties locally, due to the formation of a highly recrystallised microstructure, have

shown even poorer properties, indicating that grain boundary sliding plays a decisive

role in the deformation mechanisms of these alloys.

5.5.2 BLR tests

In this study, bolt load retention tests in tension and in compression have partially

confirmed the general tendency to have higher remaining load levels for Mg joints

bolted with Al fasteners [9,34,35]. Further, a decreasing remaining load with torque for

different bolt materials was also investigated and attested as reported in the literature [41,44]. As reported by many authors [9,34,35,41,44,45], the thermal expansion mismatch

between joint and bolt material contributes significantly to relaxation, becoming even

critical as the differences in thermal properties increase. Since the bolt and the

component to be joined have different thermal expansion coefficients, during BLR

tests, the bolt restricts the Mg from expanding freely and imposes higher stresses in

the assembly. Therefore, the closer the thermal properties are, the higher the

remaining load will be (see Figure 2.6). Furthermore, at higher preloads, the loss of

load is increased, due to the higher compressive stresses in the assembly, which

results in an increase in creep deformation of the Mg parts.

In both tensile and compressive BLR tests (see Section 4.4.2), reinforced joints have

always shown intermediate remaining load values, i.e. superior to those of AZ91D-T6

and inferior to those of MRI230D/BM. Additionally, remaining load levels of joints

welded with 8 mm tip diameter studs presented a clear advantage over those welded

with 6 mm studs. This trend can be explained by the fact that the final state of load

relaxation occurs when the stresses surrounding the bolt become less localised [38,42].

With the 8 mm welds, the stress field can, due to the larger reinforced area,

propagate to a larger extent while remaining inside of the high creep resistance

inserted material. Furthermore, Mg joints bolted with Al fasteners presented generally

the same tendency to that observed with steel bolts. However, reinforced joints

5-161

presented on average more expressive results, with remaining load values up to

97 % of those from MRI230D/BM (see Figure 4.34). Optical microscope evaluation

performed within the welded area after BLR tests revealed microstructures with none

of the discontinuities or defects that could be induced by loads imposed during the

tests.

For BLR tests in compression, the dependence of BLR behaviour on temperature

and preload force is an important issue and should always be considered during high

temperature tests of Mg alloys. Initial trials, where samples were tested at 125 °C

with Al and steel screws, presented minor differences concerning remaining loads

within this configuration. In these cases, no clear tendency could be observed

involving remaining load between reinforced or non-reinforced material (BM and

welded joints) or between different material reinforcements. This conflicts with the

literature [180], which states that moderate creep resistant AZ91D alloy is prone to

excessive creep deformation when exposed to even low levels of load at

temperatures above 100 °C. However, Table 4.19 points out a significantly low value

for remaining loads in tests carried out at 175 °C after 100 hours. Okechukwu [40]

reported also PF values (final load retained by the couple after returning to room

temperature) around zero for MRI153 and MRI230D/BM tested at 175 °C and

attributed the poor high temperature resistance of these alloys to the enlarged

Mg17Al12 grain boundary phase at higher temperatures, which is unstable at

temperatures above 120 °C. Moreover, remaining loads of up to 70 % for AE42 at

175 °C have been published in the literature [41,43,44], although under conditions

different to those employed in this work. The AZ91D welded joint was also evaluated

and clearly showed the lowest remaining load. The broad re-precipitation of β-phase

along the grain boundaries, associated with its rupture inside the grains, seemed to

facilitate grain boundary sliding, thereby decreasing creep resistance and

consequently the remaining load for this combination.

According to Figure 4.36, the joint bolted with an Al fastener presented an untypically

low remaining load of only 27 %, which is contrary to that reported in the literature [9,34,35]. BLR tests experience, during the initial period, a short load increase caused

by the thermal expansion mismatch between the Mg sample and the steel bolt. The

remaining load after testing is directly influenced by the initial load increase, which is

5-162

lower the more similar the couple and bolt materials are, i.e. similar couple and bolt

materials causes lower load increases during initial heating and therefore higher

remaining loads after the temperature cycle. Further BLR tests under these

conditions should be performed to clarify the unexpected behaviour of Mg samples

clamped with Al bolts.

Considering the results obtained from tensile and compressive BLR tests, one can

confirm the feasibility of the process to reinforce Mg components locally. In both

cases, the remaining load in welded joints exceeded that of the AZ91D BM,

indicating a promising future for the technology proposed within this study.

6-163

6 SUMMARY AND CONCLUSIONS

Mechanisms of joint formation as well as metallurgical and mechanical properties of

similar and dissimilar welds produced by Friction Hydro Pillar Processing (FHPP)

were investigated, determined and assessed. The results and analyses obtained

from these investigations lead to the following conclusions:

– In spite of it being a new welding technique, FHPP has demonstrated its suitability

for joining different magnesium (Mg) materials. The selected combination of base

materials was successfully joined in similar and dissimilar configurations using two

different geometries and a large combination of welding parameters. Both geometries

exhibited a high performance with good mechanical joint efficiency compared with

the base material within the investigated range, i.e. there is a wide operating window

for the production of satisfactory joints, considering their mechanical properties.

– A physical description of the process in terms of microstructural evolution

supported by temperature results was proposed. Based on that, the bonding

mechanisms for similar and dissimilar configurations were identified and determined,

whose behaviour accorded with the microstructural changes observed during

metallurgical characterisation.

– Welding equipment showed a high reproducibility, with a relatively low standard

deviation within the investigated range of parameters. Since the stability of the

process was proven, a further investigation of welding parameters was performed in

order to establish a correlation between the process, the final microstructure and

properties of the welded joints. In this context, the theoretical influence of the main

important process parameters on the final welded joint was confirmed by

experimental observations. The influence of those parameters on variables is

presented in Section 4. The summarised conclusions for each variable studied in

these joined materials combinations are presented below:

6-164

I. SIMILAR AZ91D-T6

Joint formation in similar AZ91D-T6 welds is characterised by the coalescence of the

plasticised layer during the initial stages of the process, followed by the formation of

a series of adiabatic shear planes with a refined dynamically recrystallised structure.

The consumable is fully plasticised across the bore of the hole and the thickness of

the work piece.

The microstructure within the recrystallised zone is characterised by refined equiaxed

grains, which tends to increase with increasing upsetting, with a complete rupture of

the lamellae β-phase. However, this phase is re-precipitated along the grain

boundaries as a probable indication of local melting. Coarse AlMn and MgSi

precipitates are not fragmented and remain in the composition of the microstructure

of the extruded material. In a region around the final frictional plane in the upper

welding area, the microstructure shows complex intercalated flow patterns, in which

part of the former base material is recrystallised in some areas, while α-matrix islands

without lamellae Mg17Al12 eutectics can be observed in between.

Maximum temperatures slightly decrease with pressure. Increasing welding pressure

and upsetting slightly enhances the mechanical properties, achieving values similar

to those from base material. Hardness profiles indicated a reduced scattering, as a

consequence of the β-phase dissolution, with no hardness loss within the

recrystallised zone. Welding pressure and upsetting were shown not to influence

hardness values significantly.

II. DISSIMILAR AE42/AZ91D-T6

In dissimilar AE/AZ welds, joint formation was characterised by different bonding

mechanisms. Within this combination, the stud yields and occupies the cavity before

a plasticised layer is created, i.e. the consumable is inserted into the cavity at a rate

faster than that at which the plasticised material develops. The consumable is

therefore neither plasticised across the bore of the hole, nor through the height of the

cavity. The joint formation can be defined as a variation of the Friction Taper Plug

Welding (FTPW) process inside a non-through thickness hole.

6-165

The microstructure of the recrystallised zone in regions at the vicinity of the joint

indicated a complete rupture of the coarse Aluminium-Rare Earth eutectics. In some

locations in the upper stud areas, clusters of particles over the bonding line were

observed. In regions towards the consumable centre, the microstructure seems to be

unaffected by the welding process to some extent. In these regions, the typical

dendritic structure with its lamellar second phase compounds remains almost

unaltered, which suggests that the consumable was not fully plasticised through its

diameter.

Mechanical tests showed the extruded material with hardness values up to 70 HV,

which indicates an overmatching condition of the extruded material in relation to the

AE42 BM. Tensile and pull-out tests revealed strengths similar or in some cases

even higher than those of the AZ91 base material.

III. DISSIMILAR MRI230D/AZ91D-T6

Within this combination, joint formation follows the same sequence discussed for

similar AZ welds. Due to the matching correlation of material/welding conditions, the

plasticised layer is generated through the consumable diameter and rises along the

hole, leaving behind a dynamically recrystallised deposited material. No bonding

formation changes were observed with increasing upsetting.

The microstructure on the MRI side of the weld is dominated by equiaxed

dynamically recrystallised grains. In particular regions of both the vicinity of the joint

and the middle of the consumable, a complete degradation of the typical base

material second phase particles Mg(Al,Ca) was observed. Additionally, dissimilar

MRI/AZ welds show a severely deformed area in the vicinity of the joint on the AZ91

side of the weld, where dynamically recrystallised grains appear in an extension from

100 μm to 400 μm from the weld line, without the presence of the eutectic structure.

In comparison with the BM, where grain size analyses indicate values ranging from

80 μm to 120 μm, in the deposit, grain size varies generally from a finer structure at

the bottom, with an average grain diameter ranging from 5 μm to 13 μm, to a coarser

structure at the top, where values varying from 18 μm to 22 μm were identified.

6-166

Mechanical properties of the welded joints showed values similar to those of the

AZ91 BM. An increased upsetting indicates no clear variation in tensile strength.

Hardness values achieve values up to 80 HV at some points, which means that in the

extruded MRI230D material an overmatching condition was created.

IV. JOINT PERFORMANCE – SIMILAR AND DISSIMILAR CONFIGURATIONS

Performance tests confirmed the feasibility of the process to locally reinforce Mg

based components. The strength of the reinforced joint relative to the BM in both

tensile and compressive BLR tests was elevated, achieving in some cases 93 % of

base material values. The higher efficiency of joints bolted with Al fasteners could not

be confirmed, although a few tests were performed under this configuration.

Creep properties were demonstrated to be inferior within the extruded material in

relation to reference base material samples. Possible formation of localised β-phase

with different morphology and the complete disruption of the second phase

intermetallic network, as a consequence of the microstructural changes imposed by

the welding process, were attributed as the most plausible reason for the poor creep

resistance on these regions. Nonetheless, the creep resistance of reinforced samples

were significantly superior to those of pure unreinforced AZ91D-T6.

The principal motivation behind this study was to perform a systematic analysis of

FHPP in Mg alloys, to understand the metallurgical phenomena and the consequent

microstructural changes imposed by the welding process. As a second issue, the

possibility of using the process to encourage a more widespread application of Mg

based materials by industry was verified. The proposed methods and mechanisms,

as well as the results of the joining parameters’ influence on temperature

development, joining mechanisms and mechanical performance of FHPP welded

joints in commercial magnesium materials, provides an important pathway to further

exploitation of scientific and technical fields of this new joining technique. Although

joint features and joining mechanisms may vary slightly within different joining

partners, owing to individual structural, thermo-physical and mechanical properties,

the results obtained from this work can be extended to other combinations of

lightweight alloys.

6-167

As summarised in Table 6.1, the establishment of the bonding mechanisms and their

influences on microstructural and mechanical properties of similar and dissimilar

welds was successfully achieved. The experimental methodology adopted at the

onset of this work and the systematic approach, in which the scientific challenges of

the subject matter have been addressed, also proved to be successful. The new and

alternative joining technique developed, characterised and demonstrated in this work

for magnesium materials, proved itself to be an interesting alternative for similar and

dissimilar configurations. Furthermore, FHPP flexibility, short joining cycles, reduced

process steps, simple machinery, due to the process robustness and possible robotic

applications, suggest potential cost savings compared to the current state of the art.

In addition, the absence of the need to develop new welding equipment, since friction

welding machines are commercially available, and the possibility to adapt industrial

machinery for FHPP make this new technology highly transferable for industrial

applications.

6-16

8

Tab

le 6

.1: S

umm

ary

of th

e w

ork.

7-169

7 SUGGESTIONS FOR FURTHER WORK

After the completion of this study, some points are recommended for further work.

– A supplementary study, focusing on creep properties and bolt load retention of

joints reinforced with different materials, would provide further design information on

joint durability and performance, which would help to accelerate the transition from

laboratory to industrial scale.

– Further mechanical and metallurgical characterisation of welded joints by means of

diverse analytical techniques, such as nanoidentation testing and TEM, would

provide a better understanding of the joining mechanisms. Such characterisation

would also be meaningful for precise characterisation of local mechanical behaviour

and/or detailed investigation of microstructural zones. Changes to fine precipitates

imposed by the welding process and their influence on the performance of the joint

could be further understood.

– With the bonding mechanism in mind, a further development of the welding

machine, incorporating a torque measurement system, would help the formulation of

the FHPP process description based on energy input. With a torque measurement

system integrated into the machine and using statistical methods, the quality of a

joint could be monitored on-line as a further stage.

– Since the first approach involving the application of FHPP in Mg alloys has been

initiated, an investigation of the joinability of high pressure die casting (HPDC) as well

as of new advanced lightweight materials, such as MMCs would be very meaningful.

Furthermore, the application of the proposed reinforcement method in real

components, such as existing gearbox housings, would significantly increase the

technical added value of the process.

8-170

8 REFERENCES

1 MORDIKE B.L. and Ebert T., 2001, “Magnesium: Properties, Applications and Potential“, Materials

Science and Engineering A302, pp.: 37-45. 2 KAINER K.U. and von Buch F., 2003, “The Current State of Technology and Potential for Further

Development of Magnesium Applications”, in Magnesium alloys and Technologies, WILEY-VCH

Verlag GmbH & CO. KG aA, Weinheim, pp. 1-22. 3 PARK S.H.C., Sato Y.S. and Kokawa H., 2003, “Effect of Microtexture on Fracture Location in

Friction Stir Welding of Mg alloy AZ61 During Tensile Test”, in Scripta Materialia 49, pp.:161-166. 4 EDGAR R.L., 2006, “Global Overview on Demand and Applications for Magnesium Alloys”,

published in Proceedings of the 7th International Conference on Magnesium Alloys and Their

Application, pp.: 3-8, 06-09 November, Dresden, Germany. 5 JOHNSON R. and Threadgill P., 2003, “Friction Stir Welding of Magnesium Alloys”, TMS,

Magnesium Technology 2003 – Edited by Howard I. Kaplan. 6 FISCHERSWORRING-BUNK A., 2005, „Das Neue BMW Aluminium-Magnesium-

Verbundkurbelgehäuse“, DGM-Fortbildungsseminar, GKSS-Forschungszentrum, 22.-24. November. 7 HOWARD-SMITH J., Teksid S.p.A., 2001, “Magnesium – An Overview”, Australian Journal of Mining. 8 AGHION E. and Eliezer D., 2004, “Magnesium Alloys - Science, Technology and Applications”,

published by The Israeli Consortium for the Development of Magnesium Technology, Israel. 9 PFISTER U., 2005, „Schraubverbindungen und Schraubtechnologie in der Powertrain-

Serienmontage“, Bad Nauheim / Frankfurt, 7. bis 9. März. 10 LEITNER H., Godor I., Eichlseder W. und Hinteregger, 2005, „Betriebsfeste Auslegung von

Schraubverbindung im Leichtmetallguss“, Bad Nauheim 17. 18. November. 11 DITZE A. and Scharf C., 2003, “Recycling of Magnesium Alloys”, in Magnesium alloys and

Technologies, WILEY-VCH Verlag GmbH & CO. KG aA, Weinheim, pp. 254-278. 12 JASCHIK C., Haferkamp H. and Niemeyer M., 2000, “New Magnesium Alloys”, in Proceedings of

the conference Magnesium alloys and their applications, 26-28 September, München, pp. 41-46. 13 CHUKO W.L. and Gould J.E., 2002, “Magnesium Joining for Automotive Applications”, EWI

Research Report No. MR0217, Edison Welding Institute, Columbus, OH. 14 ASM Specialty Handbook: Magnesium and Magnesium alloys; Ed. M.M. Avedesian ASM

International (1999). 15 http://www.efunda.com. 16 http://www.matweb.com. 17 KING J.F., 2000, “Development of Practical High Temperature Magnesium Casting Alloys”, in

Proceedings of the conference Magnesium alloys and their applications, 26-28 September, München,

pp. 14-22. 18 V. BUCH F., Schumann S., Aghion E., Bronfin B. and Mordike B.L., 2000, “Development of a Low-

cost, Temperature and Creep Resistant Magnesium Die-Casting Alloy”, in Proceedings of the

conference Magnesium alloys and their applications, 26-28 September, München, pp. 23-28.

8-171

19 PETTERSEN, K., Westengen, H., Skar, J.I., Videm, M. and Wie, L.Y., 2000, “Creep Resistance Mg

Alloy Development”, in Proceedings of the conference Magnesium alloys and their applications, 26-28

September, München, pp. 29-34. 20 MORDIKE, B.L., von Buch, F., 2000, “Development of High Temperature Creep Resistant Alloy”, in


pp. 35-40. 21 V. BUCH F. and Mordike B.L., 2000, “High-Temperature Properties of Magnesium Alloys”, in


pp. 106-129. 22 CELOTTO, S., 2000, “TEM Study of Continuous Precipitation in Mg-9 Wt% Al-1 Wt% Zn Alloy”, in

Acta Materialia 48(2000), pp. 1775-1787. 23 CHUNG, Y. and Shin, K.S., 2005, “Effects of Alloying Elements on the Microstructure and High-

Temperature Mechanical Properties of Mg-Al Alloys”, in Magnesium Technology. 24 SCHAFFER, P.L., Lee, Y.C. and Dahle, A.K., 2001, “The Effect of Aluminium Content and Grain

Refinement on Porosity Formation in Mg-Al Alloys”, Magnesium Technology – TMS 2001. 25 NAVE, M.D., Dahle A.K. and StJohn, D.H., 2000, “The Role of Zinc in the Eutetic Solidification of

Magnesium-Aluminium-Zinc Alloys”, in Magnesium Technology – TMS 2000. 26 WEI L.Y., Dunlop G.L. and Westengen H., 1996, “Development of microstructure in cast Mg-Al-Rare

Earth Alloys”, in Materials science and technology, Vol. 12, September, pp. 741-750. 27 DIERINGA H., 2006, “Vergleichende Untersuchungen zum Zug- und Druckkriechverhalten der

verstärkten und unverstärkten Magnesiumlegierung AE42”, genehmigte Dissertation, Technische

Universität Hamburg-Harburg. 28 DIERINGA H., Bowles A., Hort N. and Kainer K.U., 2005, “Microstructural Development in Tension

and Compression Creep of Magnesium Alloy AE42”, in Materials Science Forum, Vol. 482, pp. 271-

274, Switzerland. 29 POWELL, B.R., V. Rezhets, M.P. Balogh and R.A. Waldo, 2002, “Microstructure and Creep

Behaviour in AE42 Magnesium Die-Casting Alloy”, JOM August 2002, p. 34-38. 30 “MRI 153 – New Magnesium Alloy for high Temperature Applications”, DSM - Dead Sea Magnesium

Prospect. 31 “Magnesium Alloys for high Temperature Applications”, DSM - Dead Sea Magnesium Prospect. 32 EBEL-WOLF B., Walther F. and Eifler D., 2006, “Microstructure based characterization of the cyclic

deformation behaviour of the magnesium die-cast Alloys MRI153M and MRI230D”, in Proceedings of

the 7th International Conference on Magnesium Alloys and Their Applications, 06-09 November,

Dresden, Germany. 33 http://www.lam.mw.tu-muenchen.de/Lehre/Download/WfMuA/5_Magnesium.pdf. 34 WESTPHAL, K., Mulherkar, T. and Scheiding, W., 2005, “Aluminium Fasteners for Magnesium

Components”, Fastener Technological International, August. 35 WESTPHAL, K., 2003, “Verschraubung von Magnesiumkomponenten”, in Beitrag zur IIR

Fachkonferenz „Magnesium im Fahrzeugbau“, April, Marriot-Hotel Sindelfingen.

8-172

36 AUNE, T.K. and Ruden, T., 1992, “High Temperature Properties of Mg die Casting Alloys”, SAE

Technical paper, 920070. 37 SHEN G. and XU S., 2007, “Finite Element Simulation of Bolted Joints and Magnesium Bolt-Load

Retention Behaviour”, SAE Technical paper, 2007-01-1032. 38 XU S., Sohn, K.Y., Dewhirst, D. and Allison, J.E., 2000, “Finite Element Modeling of Bolt Load

Retention of Die –Cast Magnesium”, SAE Technical paper, 2000-01-1121. 39 BERTILSSON, I., Fischersworring-Bunk, A., Marx, T., Bergman, F., Kunst, M., Thiele, J. and Tan,

Z., 2006, “Aluminium Bolts in Magnesium Engine Components”, Proceedings of the 7th International

Conference on Magnesium Alloys and Their Applications, 06-09 November, Dresden, Germany. 40 OKECHUKWU, A., 2006, “Investigation of Bolt Load Retention Behaviour of Mg-Alloys”, Masters

Thesis - Technische Universität Hamburg Harburg (TUHH), Hamburg. 41 CHEN, F.C., Jones, J.W., McGinn, T.A., Kearns, J.E., Nielsen, A.J. and Allison, J.E., 1997, “Bolt-

Load Retention and Creep of Die-Cast Magnesium Alloys”, SAE Technical paper, 970325. 42 SOHN, K.Y., Jones, W., Berkmortel, J., Hu, H. and Allison, J.E., 2000, “Creep and Bolt Load

Retention Behaviour of die cast Magnesium Alloys for High Temperature Applications”, SAE Technical

paper, 2000-01-1120. 43 LUO, A. and Shinoda, T., 1998, “A new magnesium alloy for automotive powertrain applications”,

SAE Technical paper, 980086. 44 MORENO I.P., Sohn K.Y. and Jones J.W., 2001, “Bolt-Load Retention Behaviour of a Die Cast

Magnesium Rare Earth Alloy”, SAE Technical paper, 2001-01-0425. 45 SOHN, K.Y., Yurko, J.A., Jones, J.W., Kearns, J.E. and Allison, J.E., 1998, “Bolt-Load Retention

Behaviour of Die-Cast AZ91D and AE42 Magnesium”, SAE Technical paper, 980090. 46 DRUSCHITZ A.P. and Showalter E.R., 2003, “Bolt Load Compressive Stress Retention Testing of

Magnesium Alloys”, SAE Technical paper, 2003-01-0187, SAE World Congress, Detroit, Michigan,

March 3-6. 47 CROSSLAND, B., 1971, "Friction Welding - Recommended Practices for Friction Welding". in

Contemporary Physics, 12(6), pp. 559 - 574. 48 KLOPSTOCK, H. and Neelands, A.R., 1941, "An improved Method of Joining or Welding Metals",

Patent nº. GB - 00572789. 49 NCT Friction Welding website: http://www.nctfrictionwelding.com/process.php, 2005. 50 VILL, V.I., 1962, "Friction Welding of Metals", Reinhold Publishing Corporation, American Welding

Society, Inc, New York. 51 GREWE, K., 1997, “The Friction Welding Takes on New Applications”, Welding Journal, September,

pp. 39-40. 52 http://pdf.directindustry.com/pdf/thompson-friction-welding/friction-welding-applications-26030-

8501.html 53 http://www.mtiwelding.com/friction-welding-applications.cfm 54 HAZLET, T.H., 1981, “Fundamentals of Friction Welding”, Source Book on Innovative Welding

Processes, American Society for Metals, pp. 11-36.

8-173

55 ELMER, J. W. and Kautz, D., 1993, “Fundamentals of Friction Welding”, in Metal Handbook, vol. 6,

American Society for Metals, pp. 150-155. 56 BOWDEN, F.P. and Tabor, D., 1954, “The friction and Lubrication of the Solids”, Part 1, Oxford

University Press. 57 GODDARD, J. and Wilman, H., 1962, “A theory of Friction and Wear during the Abrasion of Metals”,

Wear, Vol. 5. 58 AWS Welding Handbook, 8th ed., Vol.2, American Welding Society, 1991, p. 739-763. 59 Recommended Practices for Friction welding, ANSI/AWS C6.1-89, American Welding Society,

1989. 60 ELLIS, C. R. G., 1972, "Continuous Drive Friction Welding of Mild Steel", in Welding Journal: pp.

183s - 197s. 61 LUCAS, W., 1973 "Process Parameters and Friction Welds", Metal Construction and British Welding

Journal - 5, pp.: 293-297. 62 BÖHME, D. and Hermann, F.-D., 1992, "DVS - Handbuch der Schweißverfahren Teil II:

Autogenverfahren - Thermisches Schneiden - Elektronenstrahl- und Laserstrahlschweißen, Reib-,

Ultraschall- und Diffusionsschweißen" DVS Verlag, Fachbuchreihe Schweißtechnik. 63 SPINDLER, D. E., 1994, "What Industry Needs to Know about Friction Welding", in Welding Journal,

March, pp.: 37-42. 64 DENNIN, G., 1979, "Optimierung von Einstellwerten für das Reibschweißen mit kontinuierlichem

Antrieb", in Schweißen und Schneiden, 31(7), pp. 283-289. 65 MAZAC, K., 2000, "Parameterfindung und Parameterüberwachung beim Reibschweißen", in 10.

Erfahrungsaustausch Reibschweißen, SLV München, Germany. 66 BETHLEHEM, W., 1984 (heft 01), "Geeignet zur Prozessüberwachung: Störungen des

Momentverlaufes lassen fehlerhafte Reibschweißungen erkennen", in Schweißen und Schneiden,

36(01): pp. 23-28. 67 DUFFIN F. D. and BAHRANI A.S., 1973, "Friction welding of mild steel", Metal Construction and

British Welding Journal, 05(04): pp. 125 - 132. 68 SERGIN, S.A. and SABABTSEV, V.P., 1977, "The Distribution of Temperature along the length of a

bar in relation to the speed of rotation in the friction welding of steels", in Welding Production (01): pp.

37-39. 69 VOINOV, V.G., 1968, "Mechanism of Joint Formation in Friction Welding", in Welding Production,

15: pp. 8 - 13. 70 EICHHORN F. and SCHAEFER R., 1968, "Beitrag zu den Vorgängen an der Verbindungsstelle

beim konventionellen Reibschweißen von Stahl", in Schweißen und Schneiden, 20(11): pp. 563-570. 71 HOFMANN, W. and Thome, L., 1965, "Einige Einflußgrößen auf das Reibungsschweißen von Stahl

und von Aluminium", in Schweißen und Schneiden, 17(05): pp. 194 - 199. 72 DREWS P. and Schmidt J., 1978, "Steuerung des Reibschweißvorganges zum Erzielen eines

vorbestimmten Endmaßes", in Zeitschrift für industrielle Fertigung, Springer Verlag, 68: pp. 387-393.

8-174

73 PINHEIRO, G.A., Pankiewicz, C.G., dos Santos, J.F. and Kainer K.U., 2006, “Rotational Friction

Welding of a high Creep Resistance Magnesium Alloy”, in Proceedings of the 7th International

Conference on Magnesium Alloys and Their Applications, Dresden, November. 74 PETRUCCI L. G., 1978, "Temperature distribution in friction welding", in General Engineer, July and

August, pp. 178-184. 75 SUGA Y., Miyakawa S. and Ogawa K., 1999, “Estimation of temperature distribution in the friction

weld of carbon steel by the finite element method”, in Welding International, 13(4): pp. 262 - 269. 76 TUMULURU M.D., 1984, "A Parametric Study of Inertia Friction Welding for Low Alloy Steel Pipes",

Welding Journal, Welding Research Supplement. September: pp. 289s - 294s. 77 PINHEIRO, G.A., Pankiewicz, C.G., Weis Olea, C.A., Dos Santos, J.F. and Kainer, K.U., 2007,

“Effects of Welding Conditions on Microstructural Transformations and Mechanical Properties in

AE42HP Friction Welded Joints”, presented and published on the III Workshop on Recent Advances in

Friction Welding and Allied Processes, 3/4 July 2007, Dubrovnik, Croatia. 78 MEYER A., "Friction Hydro Pillar Processing - Bonding Mechanism and Properties", Fakultät für

Maschinenbau und Eletrotechnik der Technischen Universität Carolo-Wilhelmina zu Braunschweig,

2003. 79 ASM, 1993, "Metals Handbook". Vol. 6. American Society for Metals. 80 NENTWIG, A.W.E., W. Welt et al, 1991, "Analyse des Reibschweißablaufes beim

Rotationsreibschweißen von metallischen Werkstoffen gleicher und unterschiedlicher

Warmefestigkeit", DVS Bericht: 35-44. 81 NEUMANN, A. and Schober, D., 1991, "Reibschweißen von Metallen", Verlag Technik GmbH,

Berlin. 82 PAULY, D., 1999, "Process development on an experimental Friction Hydro Pillar Processing

system", Arbeitsbereich Schiffbau, Diplomarbeit - Technische Universität Hamburg-Harburg,

Hamburg. 83 ADAM, P., 1979, “Ablauf der Verbindungsbildung beim Schwungradschweißen von hochwarmfesten

Legierungen – Temperaturverlauf und Wulstbildung”, in Schweißen und Schneiden, 31(7), pp. 279-

283. 84 KREYE, H. and Wittkamp I., 1977, "Gefügeänderung und Bindemechanismus beim

Reibschweißen". in Zeitschrift für Metallkunde, 68(4): p. 253-259. 85 WICHELHAUS G., 1974, “Messen der Stoßflächentemperatur beim Reibschweißen”, in Schweißen

und Schneiden, 26(3): pp. 97-100. 86 BETHLEHEM, W., 1984 (heft 10), "Schmelzpunkt wurde nicht erreicht: Untersuchungen des

Gefüges beim Reibschweißen". in Schweißen und Schneiden, 36(10): pp. 479-483. 87 KREYE, H., 1982, "Verbindungsbildung beim Reibschweißen", in 2. Aachener Reibschweiß-

Kolloquium, 10 – 11 March, Aachen. 88 CHENG, C.J., 1962, "Transient temperature distribution during friction welding of two similar

materials in tubular form", in Welding Research Supplement, pp. 542s-550s. 89 SLUZALEC, A., 1990, "Thermal effects in Friction Welding", International Journal of Mechanical

Sciences, 32(06): pp. 467 - 478.

8-175

90 BALASUBRAMANIAN, V., Li Y, Crompton, T., Katsube, A.S. and Soboyejo, W., 1999, "A New

Friction Law for the Modelling of Continuous Drive Friction Welding: Applications to 1045 Steel Welds"

Materials and Manufacturing Processes, Vol. 14, nº 06, pp. 845 - 860. 91 NICHOLAS, E.D. and Ellis, C.R.G., 1969, "Friction Welding of mild steel - Effect of deceleration

time", The Welding Institute Report R/RB/P40/69. 92 NICHOLAS, E.D. and Ellis, C.R.G., 1968, "Equilibrium and deceleration characteristics of mild steel

friction welds", The Welding Institute Report P/31/68. 93 DUFFIN F.D. and Crossland B., 1970, "Friction welding with sudden release of the fixed

component", Proc. of Conf. on Advances in Welding Processes, Welding Institute, pp. 25. 94 THOMAS, W., Nicholas, D., Jones, S.B., Lilly, R.H., Dawes, C.J. and Dolby, R.E., 1992, “Friction

Forming”, TWI Cambridge, Patent N°.: EP 0 602 072 B1. 95 THOMAS, W. and D. Nicholas, 1993, "On trial - a new thick plate joining technique", in TWI

Connect, April. 96 THOMAS W.M. and Nicholas D., 1992, "Leading Edge: Friction Hydro Pillar Processing (FHPP)". in

TWI Connect, June 1992. 97 NICHOLAS E.D., 1995, "Friction Hydro Pillar Processing". in 11th Annual North American Welding

Research Conference - Advances in Welding Technology, 07 - 09 November, 1995. 98 PAULY, D., 1999, "Process development on an experimental Friction Hydro Pillar Processing

system", Arbeitsbereich Schiffbau, Diplomarbeit - Technische Universität Hamburg-Harburg,

Hamburg. 99 MEYER A., Pauly D., Santos J.F., Pinheiro G., Roos A., Gibson D. and Blakemore G.R., 2001,

"Considerations on Robotic Friction Stitch Welding for the Repair of Marine Structures". in 20th

International Conference on Offshore Mechanics and Arctic Engineering. June 3-8, 2001, Rio de

Janeiro, Brazil: The American Society of Mechanical Engineers. 100 GIBSON D., Meyer A., Vennemann O., Santos J.F. and Blakemore G.R., 2001, "Engineering

Applications of Friction Stitch Welding". in 20th International Conference on Offshore Mechanics and

Arctic Engineering, June 3-8, Rio de Janeiro, Brazil: The American Society of Mechanical Engineers. 101 THOMAS, W. and Nicholas D., 1997, "The need for gas shielding - Positive advantages for two

friction processes", in TWI Bulletin, Sep/Oct 1997: pp. 84 - 88. 102 PINHEIRO, G.A., 2007, „Ein neues Fügeverfahren zur Erhöhung der Gewindetragfähigkeit im

modernen Automobilleichtbau“, InnoCarBody, Bad Nauheim / Frankfurt - 2007. 103 THOMAS, W.M. and Nicholas, E.D., 1996, “Emerging friction joining technology for stainless steel

and aluminium applications”, Presented at 'Productivity beyond 2000': IIW Asian Pacific Welding

Congress, Auckland, New Zealand, February 1996. 104 THOMAS, W.M., Nicholas, E.D. and Kallee, S.W., 2001, “Friction based technologies for joining

and processing”, present at TMS Friction Stir Welding and Processing Conference, November 2001,

Indianapolis.

8-176

105 FRED DELANY, William Lucas, Wayne Thomas, Dave Howse, David Abson, Steve Mulligan, Colin

Bird, 2005, “Advanced joining processes for repair in nuclear power plants”, presented at 2005

International Forum on Welding Technologies in Energy Engineering, September 21 – 23, Shanghai,

China. 106 SAYEE RAGHUNATHAN, 2006, “Novel technologies for repair and refurbishment”, presented at

National Welding Seminar (NWS 2006) 24 - 26 November 2006, Chennai, India. 107 HOWSE D., Lucas W., and Thomas W., 2002, “Novel joining techniques for repair in the power

generation industry”, presented at EPRI Welding and Repair Technology for Power Plants

Conference, Point Clear, Alabama, USA, 26-28 June. 108 KALLEE S. and Nicholas D., 1999, “Friction and Forge Welding Processes for the Automotive

Industry”, presented at International Body Engineering Conference, Detroit, USA, 28-30 September,

(Paper No. 99-IBECC-13). 109 ANDREWS, R.E. and Mitchell, J.S., 1990, "Underwater Repair by Friction Stitch Welding", Metals

and Materials - Marine Technology, Dezember, pp. 796 - 797. 110 BEAMISH, K., 2003, “Friction taper plug welding of 10mm AA6082-T6”, TWI internal report

768/2003. 111 BRONSERT M., 2005, "Prozessentwicklung und Bestimmung der Eigenschaften einer FHPP-

Überlappverbindung", Diplomarbeit - Vorgelegt der Technischen Universität Hamburg-Harburg im

Hochschulübergreifenden Studiengang Schiffbau, Oktober. 112 BOWLES A., Hort N., Meyer A., dos Santos J.F. and Kainer K.U., 2003, “Welding of Magnesium:

Microstructural Characterization”, in the 6th International Conference on Magnesium and their

Applications, September 2003, 917-923. 113 OGAWA K., Yamaguchi H., Ochi H., Sawai T., Suga Y. and Oki Y., 2003, “Friction Welding of AZ31

Magnesium Alloy”, Welding International, 2003, 17 (11), pp.: 879-885. 114 KATO K. and Tokisue H., 1994, “Friction Welding of Magnesium Alloys”, Welding International,

08(6), 452-457. 115 DRAUGELATES U., Schram A. und Kettler C., 1998, “Laserstrahlschweißen und Reibschweißen

von Magnesiumlegierungen”, DVS-Ber., Band 194, Schweißen und Schneiden 98, Hamburg 9/1998. 116 DRAUGELATES U., Schram A., Bouaifi B. and Kettler C., 1998, Institute of Welding and Machining

(ISAF), TU-Clausthal, Germany. 117 KATO K., Okuta M., Tokisue H. and Onoda T., 1989, Weldability of Magnesium Alloys by Friction,

Lab. Eng., Shibaura - Inst. Technol., 4, pp.: 23-38. 118 DRAUGELATES U. and Schram A., 2000, “Prozessführung und Gestaltung für das Fügen

komplexer Bauteile”, Institut für Schweißtechnik und Trennende Fertigungsverfahren, Allgemeine

Angaben zum Teilprojekt B3, TU-Clausthal, Germany. 119 KATO K. and Tokisue H., 2004, “Dissimilar friction welding of aluminium alloys to other materials”,

in Welding International, 2004, 18 (11), pp.: 861-867. 120 KOHN G., Antonsson S. and Munitz A., 1999, “Friction stir welding magnesium alloys”, in TMS

Annual Meeting and Exhibition Automotive Alloys.

8-177

121 SATO Y, Hwan S., Michiuchi M., Kokawa H., 2004, “Constitutional Liquation During Dissimilar

Friction Stir Welding of Al and Mg Alloys”, Scripta Materialia 50 (2004) 1233-1236. 122 STERN A., Munitz A. and Kohn G., 2003, “Application of Welding Technologies for Joining of Mg

Alloys: Microstructure and Mechanical Properties, Magnesium Technology 2003, TMS (The Minerals,

Metals and Materials Society). 123 SOMASEKHARAN A.C. and Murr L.E., 2004, “Microstructures in Friction-Stir Welded Dissimilar

Magnesium Alloys and Magnesium Alloys to 6060-T6 Aluminium Alloy”, Materials Characterization 52

(2004) pp.:49-64. 124 XUNHONG W. and Kuaishe W., 2006, “Microstructure and Properties of Friction Stir Butt-Welded

AZ31 Magnesium Alloy”, Materials Science and Engineering A 431 (2006) 114-117. 125 www.twi.co.uk/j32k/protected/band_10/bprfws02.html 126 PINHEIRO, G.A., 2006, “Friction Hydro Pillar Processing of Lightweight Alloys: Bonding

Mechanisms and Joint Properties”, Summer School WM, Institute for Materials Research, GKSS-

Forschungszentrum, 27.-29 September. 127 PINHEIRO, G.A., 2005, DAAD-CNPq Annual Report Nr. 1, Brazil-Germany Bilateral Scientific

Cooperation. 128 LIMA, D.B.A., 2006, “Similar and Dissimilar Rotational Friction Welding of Magnesium AZ91D-T6

Magnesium alloy: Bonding Mechanisms and Joint Properties”, Diplomarbeit - GKSS Internal Report. 129 PINHEIRO, G.A., Olea, C.A., dos Santos, J.F. and Kainer, K.U., 2007, “Microstructural and

mechanical behaviour of friction welds in a high creep resistance magnesium alloy”, published in

Advanced Engineering Materials, Vol.9, Issue 9, September, pp. 757-763. 130 DIN 65582 – Wärmebehandlung von Gußstücken aus Aluminium- und Magnesiumlegierungen

(1990); pp. 1-8. 131 ASTM-B-000661-03 – Standard Practice for Heat Treatment of Magnesium Alloys (2005); pp. 1-6. 132 ASTM E 3-95, 2000, “Standard Practice for Preparation of Metallographic Specimens”, Annual

Book of ASTM Standards Vol 03.03, pp. 1-8. 133 ASTM E 340-95, 2000, “Standard Test Method for Macroetching Metals and Alloys”, Annual Book

of ASTM Standards Vol 03.03, pp. 385-395. 134 ASTM E 112-96, 2000, “Standard Test Method for Determining Average Grain Size”, Annual Book

of ASTM Standards Vol 03.03, pp. 240-263. 135 DIN EN ISO 6507, 2005, ”Metallische Werkstoffe: Härteprüfung nach Vickers“, DIN – Deutsches

Institut für Normung e.V. 136 DIN EN 1043-2, 1996, ”Härteprüfung Teil 2: Mikrohärteprüfung an Schweißverbindungen“.

European Committee for Standardization. 137 DOLLAR A. and Meade P., 2003, “Modelling Pull-out tests of Dental Implants”, Biomedical

Engineering – Applications, basis and communications, Vol. 15, N°4, August. 138 DIN 50125, 2003, ”Prüfung metallischer Werkstoffe - Zugproben“, DIN - Deutsches Institut für

Normung e.V. 139 DIN EN 10002, 1991, ”Tensile testing of metallic materials“, DIN - Deutsches Institut für Normung

e.V., Berlin.

8-178

140 BAKKE P., Westengen H., Sannes S. and Albright D., 2004, “Powertrain Components –

Opportunities for the Die Cast AE Family of Alloys”, SAE Technical Paper, 2004-01-0655. 141 XUMING S., Sohn K.Y., Dewhirst D. and Allison J.E., 2000, “Finite Element Modeling of Bolt Load

Retention of Die-Cast Magnesium”, SAE Technical paper, 2000-01-1121. 142 XU S., Williams G., Shen G., Bouchard R., Sahoo M. and Osborne R., 2006, “Bolt-Load Retention

Testing of Magnesium Alloys for Automotive Applications”, SAE Technical paper, 2006-01-0072. 143 DIN EN 20273, 1992, ”Mechanische Verbindungselemente; Durchgangslöcher für Schrauben

(ISO 273:1979)“, DIN - Deutsches Institut für Normung e. V., Berlin. 144 ASTM E 328-86, 2002, “Standard Test Method for Stress Relaxation for Materials and Structures”,

Annual Book of ASTM Standards Vol 03.01, pp. 379-390. 145 BICKFORD J.H., 1995, “An Introduction to the Design and Behaviour of Bolted Joints”, 3rd Ed., New

York, Marcel Dekker, Inc. 146 DIN EN ISO 7089, 2000, ”Flache Scheiben - Normale Reihe, Produktklasse A (ISO 7089:2000);

Deutsche Fassung EN ISO 7089:2000“, DIN - Deutsches Institut für Normung e. V., Berlin. 147 DIN EN ISO 4014 (DIN 931), 2001, ” Sechskantschrauben mit Schaft - Produktklassen A und B

(ISO 4014:1999); Deutsche Fassung EN ISO 4014“, DIN - Deutsches Institut für Normung e. V.,

Berlin. 148 Krautkramer BoltMike III catalogue, GE Inspection Technologies. 149 ASTM E1685-00, 2006, “Standard Practice for Measuring the Change in Length of Fasteners Using

the Ultrasonic Pulse-Echo Technique”, ASTM International. 150 HUANG, Y., Hort, N., Dieringa, H. and Kainer, K.U., 2004, Advanced Engineering Materials, 6(11),

883-888. 151 DIERINGA, H., Hort, N. and K.U. Kainer, 2003, Kompozyty/ Composites, 3, 275. 152 AWS Welding Handbook, 8th ed., Vol.2, American Welding Society, 1991, p. 739-763. 153 AMANCIO, S.T., 2007, “Friction Riveting: Development and Analysis of a new Joining Technique

for Polymer-Metal Multimaterials structures”, vorgelegte Dissertation, Technische Universität

Hamburg-Harburg. 154 SILVA, A.M., 2006, “An Investigation on the Structure/Property Relationships of Solid State

Welding Processes in a Titanium Matrix Composite Alloy (Ti6Al4V + 10wt.% of TiC)”, Dissertation,

Fachbereich für Ingenieurwissenschaften, Abteilung Maschinenbau, der Universität Duisburg-Essen. 155 ZETTLER, R., Lomolino, S., dos Santos, J.F., Donath, T., Beckmann, F. Lippman, T. And

Lohwasser, D., 2006, “A Study on Material Flow in FSW of AA2024-T351 and AA6056-T4 alloys”,

GKSS Internal Report. 156 SAKKINEN D.J., “Physical Metallurgy of Magnesium Die Cast Alloys”, The Dow Chemical

company. 157 BAKKE, P., Bowles, A.L. and Westengen, H., 2006, 7th International Conference on Magnesium

Alloys and their Applications, p. 53-66. 158 FUTAMATA, M. and Fuji A., 1990, “Friction welding of titanium and SUS304L austenitic stainless

steel”, in Welding International, 4(10), p. 23-28.

8-179

159 BOWDEN F.P. and Ridler K.E., 1954, “Physical Properties of Surfaces III – The surface

temperature of solids sliding”, Proceedings of the Royal Society of London, Series A, 223(1152), April. 160 BOWDEN F.P. and Thomas P.H., 1954, “The Surface Temperature of Sliding Solids”, Proceedings

of the Royal Society of London, Series A, Vol. 223(1152), April, pp. 29-40. 161 GERLICH, A., Su, P. and North, T., 2005, “Friction Stir Spot Welding of Mg-Alloys for Automotive

Applications”, Magnesium Technology 2005, TMS, p. 383-388. 162 GERLICH, A., Su, P. and North, T., 2005, “Peak Temperatures and Microstructures in Al and Mg

Alloy FSW Spot Welds”, in Welding and Joining 2005, Vol. 10, N°6, p. 647-652. 163 SATO, Y.S., Hwan, S., Park, C., Muchiuchi, M. and Kokawa, H., 2004, “Constitutional Liquation

During Similar Friction Stir Welding of Al and Mg Alloys”, Scripta Materialia, 9, p. 1233-1236. 164 NORTH, T.H., Bendzsak, G.J., Gerlich, A., Su, P. and Cingara, G., 2007, “Transient Local Melting

in AL7075-T6 Friction Stir Spot Welds”, in Materials Science Forum, Vols. 539-543, p. 3826-3831. 165 NORTH, T.H., Bendzsak, G.J., Gerlich, A., Su, P. and Maldonado C., 2005, “Understanding

Friction Welding”, IURS, Beijing, China. 166 NORTH, T.H., Bendzsak, and C.B. Smith, 2000, “Material Properties Relevant to 3D FSW

Modelling”, Proc. 2nd International Conference on Friction Welding, TWI, Gothenburg, 26-28 June. 167 CAMPBELL J., 1997, “Castings”, Butterworth-Heinemann, 261-262, Oxford, UK. 168 http://www.meridian-mag.com/jp/magnesium/datasheet.pdf 169 http://www.thermocoax.com/9_Temperature_sensors/E196-0_Thermocouple_Type_K.pdf 170 JÜTTNER, S., 1999, “Untersuchungen zum Schutzgasschweißen von Magnesiumlegierungen für

Konstruktionsbauteile im Automobilbau”, Dissertation, Fakultät für Maschinenbau und Elektrotechnik

der Technischen Universität Carolo-Wilhelmina zu Braunschweig. 171 SOMASEKHARAN, A.C. and Murr, L.E., 2004, “Microstructures in Friction-Stir Welded Dissimilar

Magnesium Alloys and Magnesium Alloys to 6061-T6 Aluminium Alloys”, Materials Characterization,

pp.: 49-64. 172 XUNHONG, W. and Kuaishe W., 2006, “Microstructure and Properties of Friction Stir Butt Welded

AZ-31 Magnesium Alloy”, Materials Science and Engineering A 431, pp.: 114-117. 173 JOHNSON, R. and Threadgill, P., 2003, “Friction Stir Welding of Magnesium Alloys”, TMS,

Magnesium Technology 2003 – Edited by Howard I. Kaplan. 174 VDI2230, 2003, “Systematic Calculation of high duty bolted joints – Joints with one cylindrical bolt”,

VDI-Handbuch Konstruktion, Gesellschaft Entwicklung Konstruktion Vertrieb, Düsseldorf. 175 DURANDET, Y., Song. W., Cordini, P., Brandt, M. and Ostendorf, A., 2006, “Effects of Strain Rate

and Filler Alloy on the Properties of Laser Welded AZ31 Alloy Sheet”, published in Proceedings of the

7th International Conference on Magnesium Alloys and Their Application, pp.: 871-876, 06-09

November, Dresden, Germany. 176 BAI J., Sun Y., Xue F., Qiang J. And Zhu T., 2007, “Effect of Al Contents on Microstructures,

Tensile and Creep Properties of Mg-Al-Sr-Ca Alloy”, Journal of alloys and Compounds, 437, pp. 247-

253. 177 “Magnesium Alloys for high Temperature Applications”, DSM - Dead Sea Magnesium Prospect.

8-180

178 HIRAI K., Somekawa H., Takigawa Y. and Higashi K., 2005, “Effects of Ca and Sr Addition on

Mechanical Properties of a Cast AZ91 Magnesium Alloy at Room and Elevated Temperature”,

Materials and Engineering A 403, pp. 276-280. 179 GUO, J.H., Chen, L.Q., Xu, Y.B. and Lian, F.Z., 2007, “Investigation of the Compressive Creep

Behaviour of AZ91D Magnesium Alloy”, Materials and Engineering A 443, pp.: 66-70. 180 ZHANG, J., Guo, Z.X., Pan, F., Li, Z. and Luo, X., 2007, “Effect of Composition on the

Microstructure and Mechanical Properties of Mg-Zn-Al Alloys”, Materials Science and Engineering A

456, pp.: 43-51. 181 SKLENICKA, V., Svoboda, M., Pahutova, M., Kucharova, K. and Langdon, T.G., 2001,

“Microstructural Processes in Creep of an AZ91 Magnesium-Based Composite and its Matrix Alloy”,

Materials and Engineering A319-321, pp.: 741-745. 182 STJOHN, D., Qian, M., Easton, M.A., Cao, P. and Hildebrand, Z., 2005, “Grain Refinement of

Magnesium Alloys”, Metallurgical and Materials Transactions A, Volume 36, number 7, Juli.

Documents

GKSS-2008-12 [9,8 MB] (PDF)